Linköping Studies in Science and Technology Dissertations, No. 1352

Study of Wired and Wireless Data Transmissions

Allan Huynh

Department of Science and Technology Linköpings University, SE-601 74 Norrköping, Sweden

Norrköping 2010

Wired and Wireless Data Transmission

A dissertation submitted to ITN, Department of Science and Technology, Linköping University, for the degree of Doctor of Technology.

ISBN: 978-91-7393-286-8 ISSN: 0345-7524

Copyright ©, 2010, Allan Huynh, unless otherwise noted. Linköping University Department of Science and Technology SE-601 74 Norrköping Sweden Printed by LiU-Tryck, Linköping 2010. Abstract i ______

ABSTRACT

The topic of this dissertation is divided into two parts where the first part presents high-speed data transmission on flexible cables and the second part presents a wireless remote monitoring and controlling system with wireless data transmission. The demand on high-speed data communications has pushed both the wired and wireless technologies to operate at higher and higher frequencies. Classic Kirchhoff’s voltage and current laws cannot be directly applied, when entering the microwave spectrum for frequency above 1 GHz. Instead, the transmission line theory should be used. Most of the wired communication products use bit-serial cables to connect devices. To transfer massive data at high speed, parallel data transfer techniques can be utilized and the speed can be increased by the number of parallel lines or cables, if the transfer rate per line or cable can be maintained. However, the lines or cables must be well-shielded so the crosstalk between them can be minimized. Differential lines can also be used to increase the data speed further compared to the single-ended lines, along with saving the power consumption and reducing the electromagnetic interference. However, characterization for differential lines is not as straight forward as for single-ended cases using standard S-parameters. Instead, mixed- mode S-parameters are needed to describe the differential-, common- and mixed-mode characteristics of the differential signal. Mixed-mode S-parameters were first introduced in 1995 and are now widely used. However, improvements of the theory can still be found to increase the accuracy of simulations and measurements, which is proposed and presented in this dissertation. The interest of wireless solution to do remote control and monitoring for cultural building has been increasing. Available solutions on the market are mostly wired and very expensive. The available wireless solutions often offer limited network size with point-to-point radio link. Furthermore, the wired solution requires operation on the building, which is not the preferred way since it will damage the historical values of cultural heritage buildings. Wireless solutions on the other hand can offer flexibility when deploying the network, i.e., operation on the building can be avoided or kept to the minimum. A platform for wireless remote monitoring and control has been established for various deployments at different cultural buildings. The platform has a modular design

ii Abstract ______to ease future improvement and expansion of the system. The platform is based on the ZigBee standard, which is an open standard, specified with wireless sensor network as focus. Three different modules have been developed. The performance has been studied and optimized. The network has been deployed at five different locations in Sweden for data collection and verification of the system stability. The remote monitoring and control functions of the developed platform have received a nomination for the Swedish Embedded Award 2010 and been demonstrated at the Scandinavia Embedded Conference 2010 in Stockholm.

Populärvetenskaplig sammanfattning iii ______

POPULÄRVETENSKAPLIG SAMMANFATTNING

Efterfrågan på snabba datakommunikationer har drivit både den trådbundna och trådlösa tekniken att arbeta vid högre och högre frekvenser. När arbetsfrekvensen går över 1 GHz kan man inte längre förlita sig på den klassiska Kirchhoffs spänning och ström lagar. Istället måste man använda sig utav den så kallade ”Transmission line theory” för att konstruera nya elektroniker och kablar. De flesta produkter som finns på marknaden idag använder sig utav trådbundna seriella kablar för att ansluta olika enheter. För att snabbt kunna överföra stora mängder data kan parallella dataöverförningstekniker utnyttjas och hastigheten kan således ökas genom att öka antalet parallella ledare. För att till fullo kunna utnyttja den parallellism måste ledaren vara väl skyddade/avskärmad så att överhörning mellan dem minimeras. Differentiella signaler kan användas för att ytterligare öka datahastigheten och med den tekniken får man även med andra fördelar som lägre energiförbrukning och minskade elektromagnetiska störningar gentemot andra kringkomponenter. Att använda sig av differentiell teknik är inte lika simpelt som vanliga singel ledare då karakterisering av differentiella ledare inte kan göras med vanliga S-parametrar. Istället måste mixed-mode S-parameter användas för att beskriva differentiella och common- mode egenskaper samt möjliggöra omvandlingar mellan de olika lägena (mode). Teorin för att beskriva mixed-mode S-parametrar presenterades för första gången år 1995 och har sedan dess blivit väl accepterad och användas idag i stor utsträckning. Men man kan fortfarande förfina mixed-mode S-parameter teorin så att precisionen för simulering och mätningar kan ökas. Om detta finns det ett förslag på hur det kan göras presenterad i denna avhandling. Under de senaste åren har marknaden för trådlösa nätverk vuxit kraftigt, men inriktningen har i huvudsak inriktats på nät med höga överföringshastigheter. Ett exempel på detta är WLAN standarder för trådlösa nätverk. En annan väl etablerad standard är Bluetooth som dock har begränsade möjligheter att ansluta sig till flera enheter och räckvidden är i normala fall begränsad till mellan 10-100 m. För applikationer som skall styras eller små mängder data skall skickas i ett stort nätverk med hög säkerhet som krav, skapades standarden ZigBee som bygger på IEEE 802.15.4 specifikationen. ZigBee standarden ger möjligheten att koppla ihop flera enheter, upp till 65000 st., till ett enda stort nätverk med hjälp av flera

iv Populärvetenskaplig sammanfattning ______nätverkstopologier. Avståndet mellan två enheter kan vara upp till 100 m med samma sändningseffekt som Bluetooth. Inom området kultursarvskonservering för att bevara det kulturella arvet har intresset för trådlös övervakning och styrning system väckts. Anledningen är att de lösningar som idag finns på marknaden är trådbundna eller är alldeles för dyra lösningar. Oftast är den trådbundna lösningen svår att motivera då man måste göra ingrepp i byggnaden, vilket inte är önskvärt då det skadar det historiska värdet av kulturarvet. Den trådlösa lösningen kan däremot erbjuda stor flexibilitet, att man kan installera ett nätverk genom att enkelt placera ut sensorer på ett godtyckligt ställe så att åtgärder på byggnaden kan minimeras eller undvikas helt. En IT-baserad fjärrövervakning och kontroll system baserad på trådlösteknik har konstruerats för att installeras på olika byggnader med högt kulturellt värde. Systemet är uppbyggd i form av olika moduler för att underlätta framtida förbättringar och utbyggnad av systemet. Det trådlösa sensornätverket är baserat på ZigBee standarden, vilken är specificerad just för trådlösa sensornätverk som fokus. Tre olika typer av moduler har utvecklats för det trådlösa sensornätverket. Nätverket är idag installerat på fem olika platser i Sverige för datainsamling och för att verifiera systemets stabilitet. Fjärrövervakning och kontrollfunktioner i den utvecklade plattformen har visats på Scandinavian Embedded Conference 2010 i Stockholm samt nominerats för Swedish Embedded Award 2010.

Acknowledgement v ______

ACKNOWLEDGEMENT

First of all I would like to express my gratitude to my supervisor Professor Shaofang Gong, for his patients, supports and for giving me the opportunity to perform this challenging and interesting research work in the research group. Furthermore, I want to thank those people at the Department of Science and Technology who in various ways have supported me in my work. Especially, I want to express my gratitude to the following persons:

People in the Communication Electronics Research Group: Dr. Adriana Serban, Gustav Knutsson, Jingcheng Zhang, Joakim Östh, Dr. Magnus Karlsson, Owais and Dr. Qin- Zhong Ye. Also former members Andreas Kingbäck and Pär Håkansson for the friendship and many discussions concerning both the job and life.

Vinnova and Swedish Energy Agency are acknowledged for financial support of this work. Leif Odselius at Micronic Laser Systems Inc., Dr. Tor Broström and Jan Holmberg at Gotland University are acknowledged for valuable inputs to the project.

Also thanks to all my dear friends for their friendship, all the laugh and support. They have really made these years a very pleasant journey.

Last but not least I would like to thank my family Josefin, Mai and her husband Cuong, Jack, Liza, Nina and John for their love and support. I would like to express my deepest gratitude to my fantastic parents Be and Muoi for their constant support regardless of what I have decided to do or not to do.

Allan Huynh, Norrköping, December 2010.

vi List of publications ______

LIST OF PUBLICATIONS

Papers included in this dissertation:

Paper 1 Allan Huynh, Shaofang Gong and Leif Odselius, “Study of High-Speed Data Transfer Utilizing Flexible and Parallel Transmission Lines”, Proceedings of International Microelectronics And Packaging Society Nordic, Törnsberg, Norway, pp. 230-234, September 2005.

Paper 2 Allan Huynh, Pär Håkansson, Shaofang Gong and Leif Odselius, “High- Speed Parallel Data Transmission Utilizing a Flex-Rigid Concept”, Proceedings of GigaHertz 2005, Uppsala, Sweden, pp. 206-209, November 2005.

Paper 3 Allan Huynh, Pär Håkansson, Shaofang Gong and Leif Odselius, “High- Speed Board-To-Board Interconnects Utilizing Flexibel Foils and Elastomeric Connectors”, Proceedings of The 8th IEEE CPMT International Conference on High Density Microsystem Design, Packaging and Component Failure Analysis, Shanghai, China, pp. 139-142, June 2006.

Paper 4 Allan Huynh, Pär Håkansson and Shaofang Gong, “Single-ended to Mixed- Mode S-Parameter Conversion for Networks with Coupled Differential Signaling”, Proceedings of The 36th European Microwave Conference 2007, Munich, Germany, pp. 238-241, October 2007.

Paper 5 Shaofang Gong, Allan Huynh, Magnus Karlsson, Adriana Serban, Owais and Joakim Östh, “Truly Differential RF and Microwave Front-End Design”, Proceedings of IEEE Wireless and Microwave Technology Conference WAMICON 2010, Florida, USA, pp. 1-5, April 2010.

Paper 6 Allan Huynh, Jingcheng Zhang, Qin-Zhong Ye and Shaofang Gong, “Wireless Remote System Monitoring for Cultural Heritage”, Sensors & Transducers Journal (ISSN 1726-5479), Vol. 118, Issue 7, pp. 1-12, July 2010.

Paper 7 Jingcheng Zhang, Allan Huynh, Qin-Zhong Ye and Shaofang Gong, “Design of the Remote Climate Control System for Cultural Buildings Utilizing ZigBee Technology”, Sensors & Transducers Journal (ISSN 1726- 5479), Vol. 118, Issue 7, pp. 13-27, July 2010.

Paper 8 Allan Huynh, Jingcheng Zhang, Qin-Zhong Ye and Shaofang Gong, “ZigBee radio with External Low-Noise Amplifier”, Sensors & Transducers Journal (ISSN 1726-5479), Vol. 114, Issue 3, pp. 184-191. March 2010.

List of publications vii ______

Paper 9 Allan Huynh, Jingcheng Zhang, Qin-Zhong Ye and Shaofang Gong, “ZigBee radio with External Power Amplifier and Low-Noise Amplifier”, Sensors & Transducers Journal (ISSN 1726-5479), Vol. 118, Issue 7, pp. 110-121, July 2010.

Paper 10 Jingcheng Zhang, Allan Huynh, Qin-Zhong Ye and Shaofang Gong, “Reliability and Latency Enhancements in a ZigBee Remote Sensing System”, Proceedings of The Fourth International Conference on Sensor Technologies and Applications (SENSORCOMM 2010), Venice/Mestre, Italy, pp. 196-202, July 2010.

The author has also been involved in the following papers and book chapters not included in this thesis.

Paper 11 Jingcheng Zhang, Allan Huynh, Qin-Zhong Ye and Shaofang Gong, “Remote Sensing System for Cultural Buildings Utilizing ZigBee Technology”, Proceedings of the 8th. International Conference on Computing, Communications and Control Technologies (CCCT 2010), Orlando, USA, pp. 71-77, April 2010.

Paper 12 Magnus Karlsson, Pär Håkansson, Allan Huynh and Shaofang Gong, “Frequency-multiplexed Inverted-F Antennas for Multi-band UWB”, Proceedings of IEEE Wireless and Microwave Technology Conference WAMICON 2006, Florida, USA, pp. FF2-1 – FF-2-3, December 2006.

Paper 13 Pär Håkansson, Allan Huynh and Shaofang Gong, “A Study of Wireless Parallel Data Transmission of Extremely High Data Rate up to 6.17 Gbps per Channel”, Proceeding of IEEE Asia-Pacific Microwave Conference 2006, Yokohama, Japan, pp. 975-978, December 2006.

Paper 14 Duxiang Wang, Allan Huynh, Pär Håkansson, Ming Li and Shaofang Gong “Study of Wideband Microstrip Correlators for Ultra-wideband Communication Systems”, Proceedings of IEEE Asia-Pacific Microwave Conference 2007, Bangkok, Thailand, pp. 2089-2092, December 2007.

Paper 15 Duxiang Wang, Allan Huynh, Pär Håkansson, Ming Li and Shaofang Gong, ”Study of Wideband Microstrip 90º 3-dB Two-Branch Coupler with Minimum Amplitude and Phase Imbalance”, Proceedings of International Conference on Microwave and Millimeter Wave Technology, Nanjing, China, pp. 116-119, April 2008.

viii List of publications ______

Book chapter 1: Allan Huynh, Magnus Karlsson and Shaofang Gong, “Mixed-mode S-parameters and Conversion Techniques”, Advanced Microwave Circuits and Systems, INTECH, pp. 1-12, April 2010, ISBN: 978-953-307-087-2.

Book chapter 2: Magnus Karlsson, Allan Huynh and Shaofang Gong, “Parallel channels using frequency multiplexing techniques”, Ultra Wideband, SCIYO, pp. 35-54, September 2010, ISBN: 978-953-307-139-8.

List of abbreviations ix ______

LIST OF ABBREVIATIONS

ADS Advance Design Systems Balun BALance-UNbalance transformer BER Bit Error Rate BPSK Binary Phase Shift Keying C Capacitance CAD Computer Aided Design CMRR Common-Mode Rejection Ratio CRC Cyclic Redundancy Check CSMA-CA Carrier Sense Multiple Access – Collision Avoidance dB Decibel DHCP Dynamic Host Configuration Protocol DUT Device Under Test EM ElectroMagnetic EMI ElectroMagnetic Interference FFD Full Function Device FR-4 Flame Resistant 4 FTDI Future Technology Devices International. G Conductance HART Highway Addressable Remote Transducer HDMI High-Definition Multimedia Interface IC Integrated Circuit IEEE Institute of Electrical and Electronics Engineers ITN Department of Science and Technology IP Internet Protocol kbps Kilo bits per second KCL Kirchhoff’s Current Law KVL Kirchhoff’s Voltage Law L Inductance LAN Local Area Network LED Light Emitting Diode LiU Linköping University

x List of abbreviations ______

LNA Low-Noise Amplifier LOS Line of Sight LQI Link Quality Indication LR-PAN Low-Rate Personal Area Network LVDS Low Voltage Differential Signaling Mbps Mega bits per second MCU Microcontroller Unit NF Noise Figure NWK Network O-QPSK Offset Quadrature Phase Shift Keying PA Power Amplifier PCB Printed Circuit Board PER Packet Error Rate QPSK Quadrature Phase Shift Keying R Resistance RF Radio Frequency RFD Reduced Function Device RO4350B Rogers Material 4350B RSSI Received Signal Strength Indicator Rx Receiver TRL Through, Reflected and Line S-Parameters Scattering Parameters SNR Signal-to-Noise Ratio SMA Sub-Miniature connector version A SMD Surface Mounted Device SoC System-on-Chip SOLT Short-Open-Load-Through Tanδ Dielectric loss, Loss tangent, Dissipation factor Tbps Tera bits per second TDR Time Domain Reflectometer TI Texas Instruments TRL Trough-Reflect-Line Tx Transmitter UART Universal Asynchronous Receiver/Transmitter

List of abbreviations xi ______

USB Universal Serial Bus UWB Ultra Wide Band VNA Vector Network Analyzer WLAN Wireless Local Area Network ZC ZigBee Coordinator ZED ZigBee End-Device ZR ZigBee Router

Contents xiii ______

Contents

Abstract ...... i Populärvetenskaplig sammanfattning ...... iii Acknowledgement...... v List of Publications ...... vi List of Abbreviations ...... ix 1 Introduction...... 1 1.1 Background and motivation...... 1 1.2 Objective ...... 2

2 High-Speed Data Transmission ...... 3 2.1 Characteristic impedance and reflection...... 4 2.2 Crosstalk and coupling...... 7 2.2.1 Odd-mode ...... 9 2.2.2 Even-mode...... 11 2.2.3 Terminations...... 11 2.3 Differential data transmission ...... 13 2.3.1 S-Parameters...... 14 2.3.2 Mixed-mode S-parameters...... 15 2.3.3 Single-ended to mixed-mode conversion ...... 17 2.3.4 Modified single-ended to mixed-mode conversion ...... 18

3 Wireless Sensor Data Transmission ...... 23 3.1 CultureBee system overview...... 24 3.2 Wireless sensor network and ZigBee...... 25 3.2.1 Network topology...... 26 3.2.2 CultureBee nodes...... 28 3.2.3 ITN sensor end-device...... 29 3.2.4 ITN module with extra PA and LNA ...... 30 3.2.5 ITN coordinator module ...... 31 3.3 Radio range...... 31 3.4 Power consumption...... 36 3.4.1 Impact by adding extra amplifiers ...... 38 3.5 Local server...... 40 xiv Contents ______

3.6 Remote monitoring and control...... 40 3.6.1 Demonstration board for remote control...... 41 3.7 Future work...... 43

4 Results...... 47 5 Summary of the Included Papers...... 51 Bibliography ...... 57 Paper I ...... 63 Study of High-Speed Data Transfer Utilizing Flexible and Parallel Transmission Lines ...... 65 I. Introduction...... 65 II. Simulation...... 66 III. Results...... 67 a. Bandwidth...... 67 b. Skin effect and surface roughness ...... 68 c. Loss tangent...... 69 d. Crosstalk ...... 71

IV. Discussions...... 72 V. Conclusions...... 73 Acknowledgement ...... 73 References ...... 73 Paper II...... 75 High-Speed Parallel Data Transmission Utilizing a Flex-Rigid Concept ...... 77 I. Introduction...... 77 II. Design and Simulation...... 78 III. Simulation Results ...... 80 a. Via-hole ...... 80 b. Flex-only and Flex-rigid simulations ...... 82

IV. Discussions...... 86 V. Conclusions...... 86 Acknowledgement ...... 87 References ...... 87

Contents xv ______

Paper III ...... 89 High-Speed Board-to-Board Interconnects Utilizing Flexible Foils and Elastomeric Connectors...... 91 I Introduction...... 91 II. Design ...... 92 a. Flex-foil cables...... 92 b. Elastomeric (Zebra) connectors...... 93 c. Test-board ...... 94

III. Measurement Set-up ...... 95 a. Time domain ...... 95 b. Frequency domain...... 95 c. Data rate...... 96

IV. Measurement Results...... 96 a. Time domain ...... 96 b. Frequency domain...... 98 c. Data rate...... 98

V. Discussions ...... 100 VI. Conclusions ...... 101 Acknowledgements ...... 101 References...... 101 Paper IV...... 105 Mixed-Mode S-Parameter Conversion for Networks with Coupled Differential Signaling...... 107 I. Introduction...... 107 II. Mixed-Mode S-Parameters with Coupling ...... 108 III. Simulations...... 111 a. Set-up ...... 111 b. Results...... 113

IV. Discussions ...... 114 V. Conclusions ...... 115 Acknowledgements ...... 116 References...... 116

xvi Contents ______

Paper V...... 119 Truly Differential RF and Microwave Front-End Design...... 121 I. Introduction...... 121 II. Truly Differential Radio Front-End ...... 122 a. Topology...... 122 b. Methodology for TDRF design ...... 123

III. Case Study RF Front-End 6-9 GHz ...... 125 a. Differential/dipole antenna...... 125 b. Differential RF filter...... 126 c. Differential matching network...... 129

IV. Discussions...... 132 V. Conclusions...... 132 Acknowledgements...... 132 References ...... 133 Paper VI ...... 135 Wireless Remote Monitoring System for Cultural Heritage...... 137 I. Introduction...... 137 II. System Overview...... 139 a. ZigBee wireless sensor network...... 139 b. Local server...... 140 c. Remote main server and monitoring ...... 141

III. Hardware Implementation ...... 141 a. ZigBee sensor modules...... 141 b. ZigBee coordinator and router...... 142

IV. Software ...... 143 a. Battery consideration on ZigBee sensor module...... 143 b. Data display on local server...... 144 c. Main server (remote data monitoring)...... 145

V. Results...... 147 a. Current consumption ...... 148

VI. Discussions...... 150 VII. Conclusions...... 151

Contents xvii ______

Acknowledgement...... 151 References...... 151 Paper VII ...... 155 Design of the Remote Climate Control System for Cultural Buildings Utilizing ZigBee Technology...... 157 I. Introduction...... 157 II. Remote Control System Overview...... 158 III. Software Design of Wireless Sensor Network ...... 160 a. Control node registration ...... 161 b. Sensor node service discovery ...... 162 c. Sensor node working state machine...... 163 d. Control node operation...... 166

IV. Software Design of the Local Server ...... 166 a. Control and sensor node registration ...... 167 b. Sensor reading information synchronization ...... 168 c. Local server command polling...... 169

V. Software Design of the Main Server...... 170 VI. Design Result Discussion ...... 173 a. System inter-dependency...... 173 b. Service extension and reuse ...... 174 c. Automatic configuration...... 174 d. Passive command polling vs. direct command forwarding ...... 174

VII. Conclusions ...... 175 Acknowledgements ...... 175 References...... 175 Paper VIII...... 179 ZigBee Radio with External Low-Noise Amplifier...... 181 I. Introduction...... 181 II. Range and Receiver Sensitivity...... 182 III. ZigBee Modules ...... 184 IV. Software ...... 185 V. Results ...... 186 xviii Contents ______

a. LNA performance...... 186 b. Outdoor measurement...... 186 c. Indoor measurement ...... 187 d. Battery lifetime ...... 189

VI. Conclusions...... 189 Acknowledgements...... 190 References ...... 190 Paper IX ...... 193 ZigBee Radio with External Power Amplifier and Low-Noise Amplifier ...... 195 I. Introduction...... 195 II. Network Topology ...... 196 III. Radio Range and Receiver Sensitivity ...... 197 IV. ZigBee Modules...... 199 V. Current Consumption Measurement Set-up...... 200 VI. Software ...... 201 VII. Results...... 202 a. PA performance...... 202 b. LNA performance...... 203 c. Outdoor radio range...... 204 d. Indoor radio range ...... 205 e. Power consumption ...... 207

VIII.Conclusions...... 208 Acknowledgements...... 209 References ...... 209 Paper X...... 213 Reliability and Latency Enhancements in a ZigBee Remote Sensing System...... 215 I. Introduction...... 215 II. System Enhancement Software Design in ZigBee Network ...... 217 a. Control and configure the network topology...... 217 Define the ZigBee network shape ...... 218

Contents xix ______

Control the ZigBee network topology by allowing /disallowing MAC association …………………………………………………………………………….219 b. Optimize the ZigBee network latency ...... 219 c. Routers restore network information from flash memory after power reset …………………………………………………………………………….220

III. System Enhancement Design in Local Server ...... 221 a. ZigBee network failure warning function...... 221 b. Software data buffer for Internet fault tolerance...... 222

IV. Latency Test Set-up...... 223 a. AODV routing discovery minimum latency measurement...... 224 b. “Follow the topology” routing method latency measurement ...... 226

V. Network Test Result...... 226 a. Network latency test result summary...... 227 b. Temperature and humidity measurement results using our monitoring system …………………………………………………………………………….228

VI. Discussions ...... 231 VII. Conclusions ...... 232 Acknowledgements ...... 233 References...... 233

Introduction 1 ______

1 Introduction

Evolution of data transmission has always been striving towards higher data transfer rate than that in the past. Wired data transmission offers a much higher data transfer rate compares to wireless solutions. However, the wireless solution can offer a superior flexibility over the wired solution when the communicating equipment is moving or rotating. Furthermore, the wireless solution can also offer a lower installation cost when the communicating devices are located from a couple of meters to a few hundred meters away from each other. However, the development of wireless sensor networks (WSN) has been toward the lower data transfer rate to reduce the power consumption. Owing to the technological advances of wireless solutions, manufacturing of small and low-cost sensor devices has become more and more interesting. A sensor device can be placed at any spot of interest to measure the ambient condition and send the measurement data to a collector for data processing. A large number of sensors in a WSN may collect information over a large area and the network installation can be deployed very flexibly at much lower cost compare to a wired solution.

1.1 Background and motivation One common trend for both wired and wireless data transmissions is that they are pushing the operating frequency to the microwave spectrum, i.e., above 1 GHz. To build a system in the microwave spectrum requires special knowledge and careful considerations during the development phase. Traditional electronics design rules and techniques must be complemented with theories and tools for microwave. Using a transmission line without impedance control causes problems between two devices connected together at high frequency, due to signal dispersion and reflections. Furthermore, the reflected signals are distorted.

2 Introduction ______

Other phenomena to be considered are the crosstalk and electromagnetic interference between the signal lines or channels, i.e., a signal transmission in one medium creates undesired effects in another medium. The minimum space between the signal lines must be considered, e.g., inserting of guard lines between the signal lines. Today’s technology development strives for system miniaturization, which makes those solutions by increasing the distance between the lines or inserting a guarding line undesirable. Instead, differential transmission lines can be utilized to overcome the crosstalk and electromagnetic interference problem. However, in traditional microwave theory, electric current and voltage are treated as single-ended, which makes the differential designs more complicated. Furthermore, when dealing with microwave signals, classical Kirchhoff’s voltage and current laws (KVL and KCL) used for circuit analysis cannot directly be applied. Instead, the transmission line theory needs to be used. With the transmission line theory, phenomena like signal reflections can be described with its origin and methods to void them can be found. Other things like connectors and connecting pads also become important, because they introduce parasitic capacitances and inductances, leading to characteristic impedance variation.

1.2 Objective The objective of this research is divided into two parts. The first part of the work is to find solutions for parallel data transfer using cables with controlled characteristic-impedance and cables for high-speed data transmission. The second part of the work is to find solutions for remote wireless sensor networks for cultural heritage buildings. An infrastructure for data storage and access is needed. The focus is for wireless system design with low-speed data transfer but long battery lifetime for the sensor devices.

High-speed data transmission 3 ______

2 High-Speed Data Transmission

The main difference between the circuit theory based on Kirchhoff’s voltage and current laws and the transmission line theory comes from the physical dimension versus the wavelength of the signals. The circuit theory applies when the wavelength is much longer than the physical dimension of the electrical circuit. On the contrary, the transmission line theory applies when the wavelength is shorter than the physical dimension of the electrical circuit. In the transmission line theory, a line can be seen as a distributed element on which the amplitude and phase of the voltage and current vary along the line. The secondary difference is the description of voltage and current; in the circuit theory the voltage and current are space-invariant, whereas in the transmission line theory the voltage and current are described as traveling waves. As a rule of thumb, when the average size of a component is more than one tenth of the wavelength, transmission line theory should be applied. The signal wavelength (λ) on a printed circuit board (PCB) can be described with the following expression [1].

v p λ = , (1) f ε r

where vp, f and εr are the phase velocity, frequency and relative permittivity of the dielectric material, respectively. Consider an electronic device operating at 1 GHz, the signal wavelength is 14.1 cm on a PCB material with εr = 4.5. With the operating frequency at 2.4 GHz, which is widely used by today’s wireless local area network (WLAN) or wireless personal area network (WPAN) devices, the signal wavelength is

4 Characteristic impedance and reflections ______

5.9 cm. This indicates that the transmission line theory should be applied when the device size is larger than 0.59 cm at 2.4 GHz.

2.1 Characteristic impedance and reflection The relationship between the voltage and current waves on the transmission lines is described by the characteristic impedance (Z0) of the line [1].

V + V − Z −== , (2) 0 I + I − where V+, V- and I+, I- describe incident and reflected voltage and current waves, respectively. Reflection coefficient (Γ0) is another important definition that describes the ratio of the reflected to the incident waves [1].

V − =Γ , (3) 0 V +

Connecting a transmission line to another transmission line or a load (ZL) will cause reflections if the characteristic impedance of the line differs from the load impedance, which is a well known effect in transmission line theory. Equation (4) shows the relationship [1].

L − ZZ 0 0 =Γ , (4) L + ZZ 0

As shown by (4), if a transmission line is connected to a load with the same impedance, no reflection occurs. In other words the incident voltage wave is completely absorbed by the receiver device. On the contrary, if ZL ≠ Z0 reflections will occur and the reflected wave will return with the same polarity if ZL > Z0 but with an inverted polarity if ZL < Z0 [1]-[3]. When using cables to connect different devices, it is not always easy to control the impedance for the whole signal path. Figure 1 shows a time domain reflectometer (TDR) measurements of a differential twisted-pair cable with connectors of SMA, SOFIX and a new type of connectors designed by Siebert [4]. Although the presented

High-speed data transmission 5 ______connector performs better than the commonly used SMA and SOFIX connectors, it only works up to a frequency of 3.6 GHz.

Figure 1. TDR measurements with nominal impedance of 100 Ω between the differential twisted-pairs with an SMA or a SOFIX-connector, or a new type connector displayed in the same figure [4].

Figure 2 shows a flexible foil cable with parallel transmission lines for board- to-board connection. The parallel transmission lines on the flexible cable (substrate) are designed for a specific characteristic impedance, in this case 50 Ω. The so-called Zebra connector is used for connecting the flex-foil cable to the board [5].

6 Characteristic impedance and reflections ______

a) flex-foil cable with parallel transmission lines b) Zebra (elastometic) connector

c) photo of a board-to-board connector using flex-rigid cable with zebra connector Figure 2. Board-to-board connection using parallel flex-rigid cable and zebra connector [5].

The connecting pads on the flex-foil cable and the Zebra connector can be modeled with equivalent lumped element model as shown in Figure 3. The values of the equivalent components are dependent of the physical size of the connecting pads and the physical size of the Zebra connector. The characteristic impedance of the connector can be designed by controlling the size of the connecting pad for a specific Zebra connector.

High-speed data transmission 7 ______

Figure 3. Equivalent lumped element circuit of connecting pads with using Zebra connector [6].

Figure 4 shows a photo of two flex-foil cables connected together using a Zebra connector and the corresponding time domain reflectormeter (TDR) measurement. It is shown that, the characteristic impedance of the connector in combination with a flex- foil cable can be well controlled, avoiding signal reflections [5]-[7].

a) Photo of two flex-foil cables connected b) TDR measurements on two 100-mm-long together using Zebra connector Flex-foil cables connected together with and without the zebra connector in between Figure 4. Photo of the concept flex-foil cable with corresponding characteristic impedance [6].

2.2 Crosstalk and coupling Crosstalk can be referred to electromagnetic interference from one signal line to another signal lines. This effect is important to consider when designing high frequency devices. The crosstalk in the transmission lines can be described in three forms:

8 Crosstalk and coupling ______

¾ Near End Crosstalk (NEXT) is interference between a pair of transmission lines measured at the same end as the transmitter. ¾ Far end crosstalk (FEXT) is interference between pair of transmission lines measured at the other end of the transmission lines from the transmitter. ¾ Alien crosstalk (AXT) is interference caused by other transmission lines routed in close proximity to the transmission line of interest.

In combination with a perfect match system, the crosstalk will cause signal distortion and the receiver devices will absorb unwanted signals as noise. Additionally, in a system where the characteristic impedances are not matched, part of the unwanted interference will be reflected back and forward in the signal paths to increase the distortion. In a system where parallel transmission lines exist, i.e., differential signaling and parallel single-ended signaling, crosstalk or also known as line-to-line coupling arises and it will cause characteristic impedance changes. The line-to-line coupling is related to the mutual inductance (Lm) and capacitance (Cm) existing between the lines. The induced crosstalk or noise can be described by (5) and (6) [3]

dI = LV driver , (5) noise m dt dV = CI driver , (6) noise m dt

where Vnoise and Inoise are the induced voltage and current noises on the adjacent line and Vdrive and Idrive are the driving voltage and current on the active line. Since both the voltage and current noises are induced by the rate of current and voltage changes, extra care is needed for high-speed applications. The coupling between the parallel lines depends firstly on the spacing between the lines and secondly on the signal pattern sent on the parallel lines. Two signal modes are defined, i.e., odd- and even-modes. The odd-mode is defined such that the driven signals in the two adjacent lines have the same amplitude but a 180º out of phase, whereas the even-mode is defined such that the driven signals in the two adjacent lines have the same amplitude in phase. Figure 5 shows the electric and magnetic field lines in the odd- and even-mode transmissions on the two parallel microstrips. As shown in Figure 5a, the odd-mode signaling causes coupling due to the electric field between the

High-speed data transmission 9 ______microstrips. While in the even-mode shown in Figure 5b, there is no direct electric coupling. Figure 5c shows that the magnetic field in the odd-mode has no coupling between the two lines whereas, as shown in Figure 5d, in the even-mode the magnetic field is coupled between the two lines.

Current into the page Current out of the page

a) electric field in odd-mode b) electric field in even-mode

c) magnetic field in odd-mode d) magnetic field in even-mode

Figure 5. Odd- and even-mode electric and magnetic fields for two parallel microstrips [3].

2.2.1 Odd-mode The voltage noises in an odd-mode and even-mode transmissions on the parallel transmission lines can be defined with (5), leading to the following equations.

dI dI LV 1 += L 2 , (7) 01 dt m dt dI dI LV 2 += L 1 , (8) 02 dt m dt

where L0 is the equivalent lumped-self-inductance in the transmission line and Lm is the mutual inductance arisen due to the coupling between the lines. Signal propagation in the odd-mode requires I1 = -I2. Substituting it into (7) and (8), (9) and (10) are obtained.

10 Crosstalk and coupling ______

dI ()−= LLV 1 , (9) 01 m dt dI ()−= LLV 2 , (10) 02 m dt

Equations (9) and (10) show that, due to crosstalk, the total inductance in the transmission lines is reduced with the mutual inductance (Lm). Similarly, the current noises in the parallel transmission lines can be redefined (6) to the following equations.

dV ( −VVd ) CI 1 += C 21 , (11) 01 dt m dt dV ( −VVd ) CI 2 += C 12 , (12) 02 dt m dt

where C0 is the equivalent lumped-capacitance between the line and ground, and Cm is the mutual capacitance between the transmission lines arisen due to the coupling between the lines. Signal propagation in odd-mode requires V1 = -V2. Substituting it into (11) and (12) results in (13) and (14).

dV ()+= 2CCI 1 , (13) 01 m dt dV ()+= 2CCI 2 , (14) 02 m dt

Equations (13) and (14) show that, in opposite to the inductance, the total capacitance increase with the mutual capacitance. The addition of mutual inductance and capacitance causes the change of the characteristic impedance (13) and phase velocity (14), leading to (15) and (16), respectively.

ω( 0 −+ LLjR m ) Z oo = , (15) ω()0 ++ 2CCjG m 1 vpo = , (16) ()(0 m 0 +− 2CCLL m )

High-speed data transmission 11 ______

where Zoo and vpo are the odd-mode impedance and phase velocity, respectively. Consequently, the total characteristic impedance in the odd-mode reduces due to the coupling or crosstalk between the parallel transmission lines and the phase velocity changes as well.

2.2.2 Even-mode

In the case of even-mode, V1 = V2 can be substituted into (7) and (8) and I1 = I2 into (11) and (12), resulting in (17) – (20).

dI ()+= LLV 1 , (17) 01 m dt dI ()+= LLV 2 , (18) 02 m dt dV = CI 1 , (19) 01 dt dV = CI 2 , (20) 02 dt

Consequently, in opposite to the odd-mode case, the even-mode wave propagation changes the even-mode impedance (Zoe) and phase velocity (vpe) as shown by (21) and (22), respectively.

ω( 0 ++ LLjR m ) Z oe = , (21) + ωCjG 0 1 vpe = , (22) ()(0 + m CLL 0 )

2.2.3 Terminations As shown by (15) and (21) the impedance varies due to the odd- and even-mode transmissions and the coupling between the transmission lines. Figure 6 shows a graph of the odd- and even-mode impedance change as a function of the spacing between two specific parallel-microstrips. If the loads connected to the parallel lines have a simple termination as used in the single-ended case, reflections will occur due to Zoo ≠ Zoe ≠ Z0. Figure 7 shows two termination configurations, i.e., Pi- or T-termination, which can terminate both the odd- and even-mode signals in coupled transmission lines.

12 Crosstalk and coupling ______

Figure 6. Variation of the odd- and even-mode impedances as a function of the spacing between two parallel microstrips [3].

V1 R1 Differential + reciever R3 R2 -

V2

a) Pi-termination

V1 R1 + R3 - Single-ended R recievers 2 - V 2 +

b) T-termination Figure 7. Termination configurations for coupled transmission lines [3].

Figure 7a shows the Pi-termination configuration. In the odd-mode transmission, i.e.,

V1 = -V2 a virtual ground can be imaginarily seen in the middle of R3 and this forces

R3/2 in parallel with R1 or R2 equal to Zoo. Since no current flows between the two transmission lines in the even-mode, i.e., V1 = V2, this makes R1 and R2 be equal to Zoe.

High-speed data transmission 13 ______

Equations (23) and (24) show the required values of the termination resistors for the Pi- termination configuration.

= 21 = ZRR oe , (23)

ZZ oooe R3 = 2 , (24) − ZZ oooe

Figure 7b shows the T-termination configuration. In the odd-mode transmission, i.e., V1

= -V2, a virtual ground can be seen between R1 and R2 and this makes R1 and R2 equal to Zoo. In the even-mode transmission, i.e., V1 = V2, no current flows between the two transmission lines. This makes R3 to be seen as two 2R3 in parallel, as illustrated in

Figure 8. This leads to the conclusion that Zoe must be equal to R1 or R2 in serial with

2R3. Equations (25) and (26) show the required values of the termination resistors needed for the T-termination configuration [3],[8].

V1 R1 2R3

R2 2R3 V2

Figure 8. Equivalent network for T-network termination in even-mode [3].

= 21 = ZRR oo , (25) 1 ()−= ZZR , (26) 3 2 oooe

2.3 Differential data transmission Differential signaling in analog circuits are an old technique that has been used for more than 50 years. However, it has been becoming important in digital circuits since the last decenniums when low voltage differential signaling (LVDS) was developed. The reason is that increasing of the clock frequency makes crosstalk and electromagnetic interference (EMI) critical problems in high-speed digital systems as mentioned in the previous sub-section. Differential signaling is a transmission method where the transmitting signals are sent in pairs with the same amplitude but the opposite phase. The main advantage

14 Differential data transmission ______with the differential signaling is that any introduced noise affects equally both the differential lines, if the lines are highly coupled together. Since only the difference between the lines is considered, the introduced noise can be rejected at the receiver device. Moreover, the generated electric and magnetic fields from the differential line pair are more localized compared to the single-ended lines. Finally, due to the ability of noise rejection, the signal swing can decrease compared to a single-ended design and thereby the power can be saved [9]. When the signal on one line is independent of the signal on an adjacent line, i.e., an uncoupled differential transmission lines, the structure is not a truly differential pair. Therefore, while designing a differential line pair, it is beneficial to start with minimizing the spacing between the transmission lines to create as strong coupling as possible [10]. After that, one can change the conductor width to obtain the desired differential impedance. In this way the coupling between the transmission lines are maximized and all the benefits from the differential signaling are utilized. However, a strong coupling between the transmission lines leads to the fact that the characteristic impedances varies a lot as shown in Figure 6. This also leads to the fact that odd- and even-mode impedances need to be considered.

2.3.1 S-Parameters Scattering parameters or S-parameters are commonly used to describe an n-port network operating at high frequencies like microwave frequencies [1], [11]-[12]. The main difference between the S-parameters and the other parameter representations lies in the fact that S-parameters describe the normalized power waves when the input and output ports are properly terminated, whereas other parameters describe voltage and current with open or short ports. The traveling waves used in the transmission line theory are defined with incident normalized power wave (an) and reflected normalized power wave (bn).

1 an = ()n + 0IZV n , (27) 2 Z0 1 bn = ()n − 0IZV n , (28) 2 Z0

High-speed data transmission 15 ______

where index n refers to a port and Z0 is the characteristic impedance at that port. Figure 9 shows a sketch of a two-port network with the normalized power wave definitions [11].

a1 a2 S b1 b2

Figure 9. S-parameters with normalized power wave definition of a two-port network.

Equation (29) shows an S-parameters expression that describes the two-port network in Figure 9. To describe an n-port network, Equation (29) can be expanded to an nxn S-parameters matrix [1], [11].

⎧b1 ⎫ ⎡ SS 1211 ⎤⎧a1 ⎫ ⎨ ⎬ = ⎢ ⎥⎨ ⎬, (29) ⎩b2 ⎭ ⎣ SS 2221 ⎦⎩a2 ⎭

2.3.2 Mixed-mode S-parameters A two-port single-ended network can be described by a 2x2 S-parameter matrix as (29). However, to describe a two-port differential-network a 4x4 S-parameter matrix is needed, since there exists a signal pair at each differential port. Figure 10 shows a sketch of the power wave definitions of a two-port differential-network, i.e., a four-port network [12].

a1 a3

b1 P1 P3 b3 DUT

a2 a4

b2 P2 P4 b4

Figure 10. Power wave definition of a differential two-port network.

16 Differential data transmission ______

Since the differential signal is in general composed of both differential- and common-mode signals, the single-ended four-port S-parameter matrix does not provide much insight information about the differential- and common-mode matching and transmission. Therefore, the mixed-mode S-parameters must be used. The differential two-port and mixed-mode S-parameters are defined by (30) [12]-[16].

⎡bd1 ⎤ ⎡⎡ dd11 SS dd12 ⎤ ⎡ dc11 SS dc12 ⎤⎤⎡ad1 ⎤ ⎢ ⎥ ⎢ ⎥⎢ ⎥ b ⎢ SS ⎥ ⎢ SS ⎥ a ⎢ d 2 ⎥ = ⎢⎣ dd 21 dd 22 ⎦ ⎣ dc21 dc22 ⎦⎥⎢ d 2 ⎥ , (30) ⎢bc1 ⎥ ⎢⎡ cd11 SS cd12 ⎤ ⎡ cc11 SS cc12 ⎤⎥⎢ac1 ⎥ ⎢ ⎥ ⎢ ⎥⎢ ⎥ ⎢ SS ⎥ ⎢ SS ⎥ ⎣bc2 ⎦ ⎣⎣ cd 21 cd 22 ⎦ ⎣ cc21 cc22 ⎦⎦⎣ac2 ⎦

where adn, acn, bdn and bcn are normalized differential-mode incident-power, common- mode incident-power, differential-mode reflected-power, and common-mode reflected- power at port n. The mixed-mode S matrix is divided into 4 sub-matrixes, where each of the sub-matrixes provides information for different transmission modes.

• Sdd sub-matrix: differential-mode S-parameters

• Sdc sub-matrix: mode conversion of common- to differential-mode waves

• Scd sub-matrix: mode conversion of differential- to common-mode waves

• Scc sub-matrix: common-mode S-parameters

With mixed-mode S-parameters, characteristics about the differential- and common- mode transmissions and conversions between differential- and common-modes can be found [12]-[16]. The mixed-mode parameters can be used to characterize a system in microwave designs, e.g., to analyze the performance of the system, with the widely used parameter common-mode rejection ratio (CMRR). CMRR is defined as a ratio of differential- mode conversion factor

S CMRR = dd , (31) S dc

The larger the CMRR, the higher the common-mode rejection level the system gets. Two other informative parameters that one can get from mixed-mode parameters are discrimination ratio (D) and exclusion ratio (E) shown by equations (32) and (33) [10].

High-speed data transmission 17 ______

S D = dd , (32) Scc S E = cd , (33) S dd

2.3.3 Single-ended to mixed-mode conversion The best way to measure the mixed-mode S-parameters is to use a four-port mixed-mode vector network analyzer (VNA). In the case where the mixed-mode S- parameters cannot directly be simulated or measured, the single-ended results can firstly be obtained and then converted into the mixed-mode S-parameters by a mathematical conversion. This section shows how it is done for a four-port network shown in Figure 10. The differential- and common-mode voltages, currents and impedances can be expressed by (34)–(36), where n is the port number illustrated in Figure 10 [12].

− −II 212 nn dn n −= VVV 212 n , I = , (34) − dn 2

n− +VV 212 n V = , = ncn − + III 212 n , (35) cn 2

Vd Vc Z oe Z d == 2Z oo , Z c == , (36) I d I c 2

The differential- and common-mode incident- and reflected-powers are expressed with the following equations [12].

1 adn = ()dn + IZV dndn , (37) 2 Z dn 1 acn = ()+ IZV cncncn , (38) 2 Z cn 1 bdn = ()dn − IZV dndn , (39) 2 Z dn 1 bcn = ()− IZV cncncn , (40) 2 Z cn

18 Differential data transmission ______where adn, acn, bdn and bcn are normalized differential-mode incident-power, common- mode incident-power, differential-mode reflected-power, and common-mode reflected- power at port n. The voltage and current at port n can be expressed by rewriting (27) and (28) to (41) and (42).

n 0 ( += baZV nn ), (41)

n 0 ( −= baZI nn ), (42)

Inserting (34)–(36) and (41)–(42) into (37)–(40) and assuming that Zoo = Zoe = Z0, (43) and (44) are obtained.

n− − aa 212 n n− + aa 212 n adn = , acn = , (43) 2 2

n− − bb 212 n n− + bb 212 n bdn = , bcn = , (44) 2 2

As shown by (43) and (44), the differential incident and reflected waves can be described by the single-ended waves. Inserting (43) and (44) into (30), (45) is obtained.

[]mm = [ ][ ][MSMS ]−1 , (45)

⎡ dd11 dd12 dc11 SSSS dc12 ⎤ ⎡ SSSS 14131211 ⎤ ⎢ SSSS ⎥ ⎢ SSSS ⎥ where []S mm = ⎢ dd 21 dd 22 dc21 dc22 ⎥ , []S = ⎢ 24232221 ⎥ and ⎢ cd11 cd12 cc11 SSSS cc12 ⎥ ⎢ SSSS 34333231 ⎥ ⎢ ⎥ ⎢ ⎥ ⎣ cd 21 cd 22 cc12 SSSS cc22 ⎦ ⎣ SSSS 44434241 ⎦ ⎡ − 0011 ⎤ ⎢ ⎥ 1 −1100 []M = ⎢ ⎥ 2 ⎢ 0011 ⎥ ⎢ ⎥ ⎣ 1100 ⎦

As shown by (45), single-ended S-parameters can be converted into mixed-mode S- parameters with the [M]-matrix [12]-[16].

2.3.4 Modified single-ended to mixed-mode conversion Notice that the conversion method showed in the subsection 2.3.3 assumes that

Zoo = Zoe = Z0. This assumption is only true if the coupling between the differential

High-speed data transmission 19 ______signals does not exist. Figure 6 clearly shows that if the coupling between the transmission lines exists then Z0 ≠ Zoo ≠ Zoe. Although this conversion method is widely used, the weakness of the conversion method has been noticed by people working in the area [3], [14] and [17].

Base on this observation, two new parameters koo and koe depending on the coupling between the transmission lines are introduced by the author. In that way, the correct differential- and common-mode impedances are included in the conversion.

Inserting Zoo = kooZ0 and Zoe = koeZ0 into (37)–(40), (46)–(49) are obtained.

()1+ ( noo − − 212 n )+ (1− )( noo − − bbkaak 212 n ) adn = , (46) 22 koo

()1 + ( noe − + 212 n )+ (1− )( noe − + bbkaak 212 n ) acn = , (47) 22 koe

()1+ ( noo − − 212 n )+ (1− )( noo − − aakbbk 212 n ) bdn = , (48) 22 koo

()1 + ( noe − + 212 n )+ (1 − )( noe − + aakbbk 212 n ) bcn = , (49) 22 k oe

Inserting (46)–(49) into (30), (47) is obtained.

mm −1 []= ()[]1 [ ]+ [ 2 ] ( [ 1 ]+ [ 2 ][SMMMSMS ]) , (50)

20 Differential data transmission ______

⎡ 1+ k 1+ k ⎤ oo − oo 00 ⎢ 22 k 22 k ⎥ ⎢ oo oo ⎥ 1+ k 1+ k ⎢ 00 oo − oo ⎥ ⎢ 22 k 22 k ⎥ where []M = ⎢ oo oo ⎥ and 1 1+ k 1+ k ⎢ oe oe 00 ⎥ ⎢ 22 k 22 k ⎥ ⎢ oe oe ⎥ 1+ k 1+ k ⎢ 00 oe oe ⎥ ⎢ ⎥ ⎣ 22 koe 22 koe ⎦

⎡ 1 − koo 1 − koo ⎤ ⎢ − 0 0 ⎥ 22 k 22 k ⎢ oo oo ⎥ ⎢ 1 − k 1 − k ⎥ 0 0 oo − oo ⎢ 22 k 22 k ⎥ []M = ⎢ oo oo ⎥ 2 1 − k 1− k ⎢ oe oe 0 0 ⎥ ⎢ 22 k 22 k ⎥ ⎢ oe oe ⎥ 1− k 1− k ⎢ 0 0 oe oe ⎥ ⎢ ⎥ ⎣ 22 koe 22 koe ⎦

As shown by (50), the single-ended S-parameter representation can be converted to mixed-mode S-parameters with the [M1] and [M2] matrixes for coupled transmission lines [18]. To include the odd- and even mode impedances, one must first find these impedance values. This can be done with simulations using a field solver or measurements using a differential time domain reflectometer (TDR). In fact, companies providing VNA can provide TDR as an embedded module, so only one apparatus is needed for real measurements. Another way to find the odd- and even- mode impedances of the transmission lines is to determine the coupling co-efficient C, according to [2]:

1− C ZkZ == Z , (51) oooo 0 1+ C 0 1+ C ZkZ == Z , (52) oeoe 0 1− C 0

Wireless sensor data transmission 23 ______

3 Wireless Sensor Data Transmission

Guglielmo Marconi journeyed from Italy to England February 1896 to show the British telegraph authorities what he had developed in the way of an operational wireless telegraph apparatus. The same year on the 2nd of June, his first British patent application was approved. In cooperation with Mr. W.H. Preece, they manage to transmit a radio signal over the air in a distance of 1.75 mile (2.81 km) in July 1896 and the timeline of wireless technology started [19]. The first topic about two-way radio conversation was covered in Boston Sunday Post. The article presented “Talking by Wireless as You Travel by Train or Motor” which was tested by Mr. Brandy and Harold J. Power. But it was not until 1980s that the technology became available for the average consumer around the world. In September 1999, Business Week announced that the network microsensor technology is one of the 21 most important technologies for the 21st century. Wireless sensor network (WSN) can form a large remote monitoring and control system, which can be deployed on the ground, in the air, on a human body, in a vehicle and inside a building. The collected data can be used for analysis, e.g., for weather forecasting or for analyzing the environmental changes. Potential applications for WSN are many such as military sensing, security, air traffic control, traffic surveillance, video surveillance, industrial and manufacturing automation, robotics, environment monitoring and building monitoring. The development of WSN requires technologies from mainly three different research areas covering: hardware, software and algorithm. Progress in each of these areas has driven the research in wireless sensor networks [20], [21]. Applying WSN to cultural heritage building is an interesting area due to the requirement of cultural heritage conservation in the world. Operations on the building should be kept to the minimum, therefore battery-powered WSN is of great advantage to use [22], [23].

24 CultureBee system overview ______

3.1 CultureBee system overview CultureBee is the name of the developed system for wireless remote monitoring of cultural buildings. The collected data is used to learn more about the climate in every building with unique cultural and historical values and also to develop control functions applied to heating and ventilation systems. Figure 11 shows an overview of the CultureBee system. From the left hand side, there are a number of WSN deployed at different interested buildings, such as churches and museums. Each WSN is connected to a local server. The local server with a so-called coordinator is the heart of the WSN. Furthermore, the local server also functions as an intermediate storage place of the acquired data from the WSN and as a gateway between the WSN and the Internet. Through the Internet the local server can automatically synchronize all the acquired data to a main server. The main server stores data synchronized from all connected local servers in a database and allows users (clients) to remotely access the data via the Internet. The CultureBee System is currently accessible at www.culturebee.se.

Figure 11. System overview of CultureBee [24].

Wireless sensor data transmission 25 ______

3.2 Wireless sensor network and ZigBee Some existing protocols for wireless networks, e.g., mobile ad hoc networks (e.g. Bluetooth), wireless local area network (WLAN) and cellular systems cannot be applied directly to WSN due to the fact that they do not focus on computation, storage and low power consumption constraints as needed for sensor nodes. To meet the WSN requirement, ZigBee has been developed as a high-level communication protocol using small and low-power digital radio based on the IEEE 802.15.4-2006 standard [25], [26]. Table 1 shows a comparison of some different standards. Apparently, ZigBee is more suitable for WSN due to longer battery lifetime and support for larger network size.

Table 1. Wireless standard comparison. ZigBee Bluetooth WLAN GSM/GPRS Application Monitoring & Cable Local data Wide area voice focus Control replacement access & data Battery lifetime 100 – 1000+ 1 – 7 0.5 – 5.0 1 – 7 (days) Network size 65 000 7 32 1 Data rate (kbps) 20 – 256 720 – 2400 11 000+ 64 – 230 Focus Low Power and Low Cost Speed and Range and Low Cost Flexibility Quality ZigBee operates in the industrial, scientific and medical (ISM) frequency bands defined in the IEEE 802.15.4 standard spread among 27 different channels divided into three sub-bands as shown in Table 2. The transmitted data is modulated with Direct Sequence Spread Spectrum (DSSS), to spread each bit of information into 2-bit (BPSK) or 4-bit (O-QPSK) symbol sequence. The spread symbols are more resistant against interference within the operating frequency band and thus improve the signal-to-noise ratio (S/N) in the receiver. Furthermore, to avoid packet collision, the standard provides the usage of Carrier Sense Multiple Access Collision Avoidance (CSMA-CA) or Guarantee Time Slot (GTS) techniques. Each node utilizing CSMA-CA must listen in the medium prior to transmit. If the energy found is higher than a specific level, the transmitter must wait for a random time interval before trying for a new transmission. The GTS technique requires that all the nodes in the system must be synchronized and each node is given a time slot to transmit data.

Table 2. IEEE 802.15.4 bands and channels. Frequency band (MHz) Channels Support Area 868.0 – 868.6 1 Europe 902.0 – 928.0 2-10 North America 2400 – 2480 11-26 Worldwide

26 Wireless sensor network and ZigBee ______

Other interesting network protocols that are suitable for WSN based on the IEEE 802.15.4 standard are WirelessHART and ISA SP100. Table 3 shows a summary of the difference between these protocols. ZigBee is selected for the CultureBee system because of the maturity of the standard at the time when the project started.

Table 3. Protocols for wireless sensor network [25], [27], [28]. ZigBee WirelessHART ISA SP100 Target market Consumer Industry Industry Target applications Monitor, control and Industrial control Process control automation Factory automation Channel Agility Hopping Hopping Hopping/Agility Topology Mesh/Star coexist Mesh Mesh, Tree Device Types FFD, RFD FFD FFD, RFD Battery lifetime Better Good Better Sleeping Routers No Yes Yes Latency 4 ms 8 ms 8 ms Encryption AES128 AES128 AES 128 Cost Lowest High Medium Message Priority No Yes Yes Certification Yes Yes Yes Program

3.2.1 Network topology The ZigBee network supports star, tree and mesh network topologies, as illustrated in Figure 12. Depending on the environment, different network topologies can be used with the cost of network complexity, reliability and signal latency. The star network is the simplest topology with point-to-point connection and has the lowest signal latency. Since the end-devices communicate directly with the coordinator in the star network, the risk for network failure is kept to the minimum. The network fails only when the coordinator fails, but the wireless network coverage is limited by the radio range between the coordinator and the end-device in the star network. The wireless network coverage can be extended with a so-called tree or mesh topology, by adding routers between the end-devices and the coordinator. The tree topology still uses point-to-point communication between the devices, which makes the connection predictable and the complexity of the network is moderate. The drawback is that if one of the routers fails, all the devices which have a signal path to the coordinator via that router will also be disconnected from the network.

Wireless sensor data transmission 27 ______

a) star topology b) tree topology c) mesh topology Figure 12. Illustration of different network topologies.

A mesh network can be used to avoid this problem by self-repair of the network, i.e., by reconnect of the disconnected devices to another neighboring router and rejoin to the network. Figure 13 shows a simple example of a mesh network for controlling a lamp where the switch is the coordinator. The circles with dots are routers placed between the switch and the lamp being the end-device. Figure 13a shows a complete ZigBee network with all router nodes functional as it should. If for some reason some nodes are broken or blocked as shown in Figure 13b marked with black dots, the network can self-repair and automatically re-configure the network to re-route the signal so that it reaches the final destination as shown in Figure 13c [25]. This self- reparation function provided by the mesh topology gives robustness to the network, but at the cost of increased complexity and latency time. However, for applications where low latency time is required but the star topology cannot provide enough range, adding extra amplifiers to increase the radio range (within the regulation limits) might be a better solution than adding routers between the communicating devices. Adding extra amplifiers can extend the radio range, i.e., increase the radio coverage. In the case of tree and mesh network topologies, the added extra range for the routers result in fewer routers needed in a network to cover a larger geographic area. However, increasing the signal output from the transmitters is not always desirable if the network is very dense. This is due to the higher risk of more signal interference and the function of CSMA-CA, which will force the nodes to delay if the transmission channel is occupied, thereby increasing the signal delay in the network.

28 Wireless sensor network and ZigBee ______

a) ZigBee Mesh network b) Mesh network with broken link

c) Self-heal mesh network Figure 13. ZigBee mesh network [25].

3.2.2 CultureBee nodes ZigBee devices are described based on two main types. The ZigBee logical device defines the roll of the device in the network, whereas the physical device decides the capacity of a device. ZigBee logical devices defined in the standard are end-device, router and coordinator

¾ ZigBee End-Device (ZED): This is the cheapest and most basic device in the ZigBee network. It only contains enough functionality to talk to other devices, which means that two ZED cannot talk to each other. The device contains an algorithm for sleep mode so that the battery lifetime can be prolonged. ¾ ZigBee Router (ZR): This device is used to extend the geographical network coverage and to route the data packets in the network. It can also be used for running simple applications and acts as a ZED. In general, the routers are expected to be active all the time and must be mains-powered according to the ZigBee specification.

Wireless sensor data transmission 29 ______

¾ ZigBee Coordinator (ZC): This device is responsible for creating the WSN. It also initiates and maintains the devices in the network and can further bridge the network to other networks. In each ZigBee network, only one ZC is allowed.

The physical device type is defined in the IEEE 802.15.4 standard with full function device (FFD) and reduced function device (RFD). A FFD in the ZigBee specification must be capable to implement the full ZigBee Stack and operate all three logic device types. RFD can on the other hand only be an end-device. A FFD can communicate with any device whereas RFD can only communicate with a FFD. The RFD is used for simple applications as switches or sensor devices with low power capacities. Figure 14 shows the WSN hardware developed at the Department of Science and Technology (ITN), Linköping University for the CultureBee system which is based on the ZigBee specification. Table 4 shows a short summary of the differences between the three modules.

a) Sensor end-device b) Router c) Coordinator Figure 14. Photo of the ITN ZigBee modules.

Table 4. Short summary of ITN ZigBee modules. ITN modules Chip Signal Output (dBm) Sensor End-device TI CC2430 -0,4 Router TI CC2430 20.0 Coordinator TI CC2530 1.0

3.2.3 ITN sensor end-device The sensor end-device used for WSN at ITN is built with the RF transceiver chip CC2430 provided by Texas Instruments. The sensor module (Figure 14a) is

30 Wireless sensor network and ZigBee ______connected with a temperature and relative humidity sensors to collect the interested ambient environment information as shown in the block diagram in Figure 15. The sensor is located outside the casing to make sure that it is the ambient data the sensor is collecting. The device is, as seen in Figure 14a, powered by a 3.6 V lithium battery (half AA size) with a capacity of 1100 mAh. The signal output is programmed to be - 0.4 dBm.

Figure 15. Block diagram of ITN ZigBee sensor end-device module.

3.2.4 ITN module with extra PA and LNA Since the router (Figure 14b) is defined to be mains-powered with the ZigBee specification, the device needs not to be optimized for low power consumption. Instead, this device is equipped with an extra power amplifier (PA) and low-noise amplifier (LNA) to increase the output power and receiver sensitivity as shown in Figure 16. The used IC is T7024 provided by Atmel [29]. The implemented PA is a three stage amplifier with an analog input control (ramp) for control of the signal output power. The same control signal can also be used to switch the PA to the power-down (standby) mode when the module is not transmitting. The maximum gain of the PA is 23 dB. Typical noise figure (NF) of the LNA is 2.1 dB at the frequency range between 2.4 and 2.5 GHz. Two extra switches are added to switch the module between transmit and receive modes [29].

Wireless sensor data transmission 31 ______

Figure 16. Block diagram of ITN ZigBee router module.

3.2.5 ITN coordinator module The coordinator (Figure 14c) uses TI CC2530 instead of TI CC2430 due to the fact that the UART implementation in TI CC2430 is not stable, when operating over a long period of time. TI CC2530 is connected with a UART-to-USB converter IC (FTDI FT232BL) for connection to the USB-port instead of a serial RS-232 port, as shown in Figure 17. This is done for the convenient to connect the coordinator to a modern computer where the RS-232 port is non-existent. The coordinator is powered from the USB-port so no extra external power supply is needed.

Figure 17. Block diagram of ITN ZigBee Coordinator module.

3.3 Radio range Wireless communication is achieved through transmission of signal energy over the air from one location to another as electromagnetic wave. Therefore, the radio

32 Radio range ______propagation can be polarized, damped, reflected and blocked. Generally, in an open space, the propagation of the radio signal can be described by the Friis transmission formula (53).

2 ⎛ λ ⎞ = GGP rtr ⎜ ⎟ Pt , (53) ⎝ 4πR ⎠

where Pt, Pr, Gt, Gr, λ and R are transmit power, receive power, antenna gain of the transmitter, antenna gain of the receiver, radio wavelength and the distance between the transmitter and receiver, respectively [2], [30], [31]. However, when verifying the radio range of the ITN ZigBee modules, the alignment and polarization of the antennas have been noticed to affect the radio range. Moreover, the signal can further be affected by the following physical properties in the geographical area with obstacles [33]:

¾ Reflection: this occurs when the radio wave meets the interface between two different media so that it returns into the medium from which it originated. Unless an incident is propagated perpendicular to the surface, the signal wave will change the direction according to the Law of reflection which says that a reflected

ray lies in the same plane of incidence and has an angle of reflection (θr) equal to

the angle of incidence (θi).

= θθ ir , (54)

The amount of reflected power is dependent of the polarization of the wave and is

described by the Fresnel equations. The reflected wave (Rs) that is perpendicular polarized to the plane illustrated in Figure 18 is described as

⎛ cos − nn cosθθ ⎞ ⎜ 1 i 2 t ⎟ Rs = ⎜ ⎟, (55) ⎝ 1 cos i + nn 2 cosθθ t ⎠

whereas reflected wave (Rp) that is parallel polarized with to the plane illustrated in Figure 18 is described by

Wireless sensor data transmission 33 ______

⎛ cos − nn cosθθ ⎞ ⎜ 1 t 2 i ⎟ RP = ⎜ ⎟ , (56) ⎝ 1 cos t + nn 2 cosθθ i ⎠

Figure 18. Variables definition for Fresnel equation [32].

¾ Refraction: this occurs, in concurrent as reflections, when the radio wave travels through a surface that separates two media. The Law of refraction says that a

refracted ray lies in the same plane of incidence and has an angle of refraction θt

that is related to the angle of incidence θi by

n2 sin t = n1 sinθθ i , (57)

where n1 and n2 is a the index of refraction. The amount of transmitted power (T) that transmits through the media is described as

T = 1− R , (58)

where R is the amount of reflected power described by Fresnel equations (55) and (56) [32]. ¾ Penetration: this is related to the penetration depth which measures how deep the electromagnetic wave can penetrate into a material. If the penetration depth is deeper than the thickness of the material, then the signal goes through it. According to Beer-Lambert Law [33], the intensity of the electromagnetic wave I(z) inside a material falls off exponentially from the surface as

−αz ( ) = 0eIzI , (59)

34 Radio range ______

where I0 is the intensity of the incident wave and z is the penetration depth. ¾ Signal multipath: in physics, interference is superposition of two or more waves that results in a stronger signal (constructive interference) or weaker signal (destructive interference). This is known as multi-path propagation and fading in wireless communication. The transmitted signal wave can travel to the receiver via different paths and cause wave interference. ¾ Signal interference: This may occur when different communication systems are operating at the same frequency. Signals from other external wireless networks will operate as noise or interference.

This leads to the fact that to predict how the signal reception will be at the receiver device, e.g., in the indoor environment will be very difficult. Furthermore, the person who deploys a new WSN in a building has to take all these factors into account in order to establish the best network. Noise is another important parameter when consider radio range. Noise figure (NF) is a measure of how much the signal-to-noise (SNR) degrades as the signal passes through a system, e.g., over the air with attenuation. If a system is noiseless, then

SNRout = SNRin, regardless of the gain or attenuation. This is because the input signal and the input noise are amplified by the same factor and no additional noise is introduced. The most commonly accepted definition of NF is [2], [34]

SNR NF = in , (60) SNRout

where SNRin and SNRout are the SNR ratio measured at the input and output of the system, respectively. The equation below shows the minimum signal power that a system can detect with the minimum SNR [34].

Pin.min [dB] = kT [dBm/Hz] + 10 log B [dB] + SNRmin [dB] + NFtot [dB], (61)

where Pin.min is the minimum detectable power with SNRmin required for a certain data rate and modulation order, k, T and B are Boltzmann’s constant, temperature and bandwidth, respectively. From (61), the first three terms are

Wireless sensor data transmission 35 ______predefined parameters by the thermal noise and the specified bandwidth used in a certain standard, which cannot be changed. Consequently, the only term left to improve the reception is NFtot. Consider a cascade of amplifier, the total noise factor (NFtot) of the system can be expressed with (62) [34]-[36].

NF2 −1 NF3 −1 tot NFNF 1 += + +L, (62) G1 GG 21

where NF1, NF2 and NF3 are the noise factor of the first, second and third stages of systems or components. G1 and G2 are the power gain from the first and second stages of the systems, respectively. As the equation shows, the noise figure of the whole system is dominated in the first few stages. Thus, to maximize the wireless transmission range, it is important that the first stage of the receiver has a low NF but high gain (G1). This indicates that to add a high quality LNA to the receiver will increase the radio range even though the transmitter keeps the same output transmit power. Radio range of the ITN ZigBee sensor module has been measured both for outdoor and indoor environments. The modules can establish a connection with zero bit error rate (BER) with a range up-to 144 m at outdoor and 68 m at indoor [37], [38]. Extra PA and LNA have been added to the ITN ZigBee router module to increase the transmit power and receiver sensitivity, in attempt to increase the radio range as described by equation (53). The outdoor radio range measurement was performed in an open area with a range of 1600 m achieved [39]. As shown in Table 5, the added PA and LNA do help to increase the radio range over the air. However, this is done at the cost of higher current consumption by the module.

Table 5. Outdoor range LOS radio range. Device Max Distance current (m) (mA) Standard ZigBee module with 3.6 V supply 34.0 144 ZigBee module with LNA activated only with 3.3 V supply 54.8 403 ZigBee module with PA activated in the transmitter and LNA 184.5 1600 activated in the receiver. Both modules are driven with 3.3 V supply

36 Wireless sensor data transmission ______

3.4 Power consumption The reason why ZigBee is a low energy protocol is due to the fact that one can put the end-device into very deep sleep-mode while keep the duty cycle really low, as illustrated in Figure 19. The level of duty cycle is dependent of the communication mode. A ZigBee network can operate in two modes: beacon mode and non-beacon mode. In the beacon mode, the ZigBee devices are periodically transmitting beacons to confirm their present to the network. Between the beacon intervals, the transceiver can be put into sleep mode to lower the duty cycle. In the non-beacon mode, data transmission is less coordinated as any device can communicate with the coordinator anytime. However, this operation can cause different devices within the network to interfere with one another. Furthermore, both the coordinator and the routers must always be awake to listen to signals, thus requiring more steady power supply. On the contrary, the end-device only activates when an external stimulus is detected and thereby keeping the duty cycle to the minimum [25].

Figure 19. Duty cycle D is describe by duration τ over the period time T.

The WSN in the Culturebee system operates in the non-beacon mode. The ITN sensor device is programmed to enter the sleep-mode when there is no task to perform and the radio part only activates when the ambient temperature change has been detected [40]. This is a proven method for reducing power consumption. The following operation states are implemented in the sensor module [41]:

¾ Sensing data ¾ Send sensor data ¾ Status check ¾ Send data request

Wireless sensor data transmission 37 ______

At the sensing data state, the module is programmed to enable the sensors and collect the data. The collected data is stored in an onboard flash memory and compared with previous stored value. If the sensor value is the same as the previous stored value, there is no need to send the data to the local server for storage. If the collected data is different from the previous stored value, the module will switch to the Send Sensor Data mode and send the newly collected data to the local database in the local server. The Status Check mode is implemented to let the end-device notify the network that it is still available. This mode is necessary due to the implementation of Sensing Data mode. If the sensed data is stable for a long time, the network will not know if the sensor device is malfunctioned, the communication is blocked or the end-device has run out of battery. At Send Data Request mode the end-device will check if there is any command to be taken from the network. Table 6 shows the duration of the different operations and two different profiles with the corresponding duty cycle. As shown in the table, the duty cycle data communication is kept very low in comparison with that for operating the sensors.

Table 6. Duration and duty cycle of ITN ZigBee modules. Module Operation Duration Period (Duty cycle) Period (Duty cycle) Data Request 8.6 ms Every 1 min (0.0143 %) Every 20 min (0.0007 %) Sensor Sensing 1380 ms Every 6 min (0.3833 %) Every 30 min (0.0767 %) Sending Data 7.8 ms Every 30 min (0.0004 %) Every 60 min (0.0002 %) Status Report 7.8 ms Every 60 min (0.0002 %) Every 60 min (0.0002 %)

The power consumption on the sensor module can further be lowed to prolong the battery lifetime by lowering the supply voltage to the radio and sensor chips or disable parts that is not need for the moment as much as possible. The switching power dissipated by a chip using static CMOS gates is given by [40], [42]

2 fCV , (63) where C is the capacitance being switched per clock cycle, V is voltage, and f is the switching frequency. As the equation shows, the power consumption is proportional to the square of the voltage. The ITN ZigBee end-device is driven by a 3.6 V lithium battery without any voltage regulator. The power consumption from the CMOS chip and the sensor can be lowered by lowering the supply voltage and the switching

38 Energy consumption ______frequency, as shown by (63). For example, if the supply voltage is lowered from 3.6 to 2.5 V, the power consumption of the sensor module can be lower by

− 22 = %7.516.3/5.21 , (64)

Table 7 shows the current consumption of the ITN ZigBee end-device module, when the transmitter is activated to send out continues wave. Although stability test of the module at lower supply voltages has not been conducted, the signal output from the transmitter looks promising.

Table 7. ITN ZigBee module current consumption with different supply voltage. Supply voltage (V) Current consumption (mA) 2.5 22 2.8 24 3.0 27 3.3 31 3.6 34

Furthermore, since TI CC2430 can operate with a 16 or 32 MHz oscillator, the lower frequency oscillator should be used as often as possible to conserve the battery lifetime, i.e., when the radio part is not active since it needs the higher frequency clock to operate. Table 8 shows a summary of which part of the module is enabled/disabled at different operation modes.

Table 8. Operation modes with different parts of the module enabled/disabled. Operation Mode Sensor Microcontroller Radio 32 MHz 32.768 kHz OSC OSC Sensing Data Enable Enable Disabled Enable Enable Send Sensor Data Disabled Enable Enable Enable Enable Status Check Disabled Enable Enable Enable Enable Send Data Disabled Enable Enable Enable Enable Request Sleep Mode Disabled Disabled Disabled Disabled Enable

3.4.1 Impact by adding extra amplifiers As shown in Table 6, the duty cycle of the ZigBee protocol can be very low. In some cases, one can increases the power consumption during the active period without affecting the battery lifetime too much if the duty cycle is kept very low relative to the power consumption by the sensors. Figure 20 and Table 9 show the current

Wireless sensor data transmission 39 ______consumption during the data request mode, which was investigated in Papers VIII and IX. The numbers listed in Figure 20 correspond to those in Table 9.

Figure 20. Voltage drop over a resistor connected in serial with the ZigBee module during data request sequence.

Table 9. Current consumption summary for data request sequence. Interval Description Standard With PA and LNA Current Duration Current Duration (mA) (ms) (mA) (ms) 1 Power saving mode 0.0008 - 0.64 - 2 Start-up @ 16 MHz 11.0 0.92 9.5 0.81 3 MCU @ 32 MHz 14.7 1.86 13.1 2.11 4 Rx 34.0 1.92 54.8 2.00 5 Rx –> Tx 22.5 0.20 39.1 0.21 6 Tx 31.6 0.56 184.5 0.63 7 Tx –> Rx 24.6 0.12 11.0 0.08 8 Rx 34.0 1.18 54.8 1.02 9 Packet processing @ 14.7 1.19 12.8 1.22 32 MHz 10 Processing sleep @ 16 9.2 0.64 7.4 0.53 MHz Total 191.4 8.59 345.7 8.61 mA*mS mA*mS

Using the extra external PA and LNA will increase the current consumption with 152.9 mA and 20.8 mA during transmit and receive mode, respectively. The table also shows that the radio active period is in total barely 4 ms. This leads to the fact that

40 Energy consumption ______adding extra PA and LNA only reduces the total battery lifetime by 4.5 % or 2.6 % when using the duty cycle profiles listed in Table 6.

3.5 Local server A local server is a simple computer, e.g., a Netbook, located at each building where the deployed wireless sensor network is located. The main task of the local server is to store all the locally sensed data into a database. The local server also serves as a gateway between the WSN and the Internet so that the local database can periodically synchronize with a remote main server. Simple tasks like adding new sensor modules and alarming local users when any system failure occurs are done by the local server. Figure 21 shows a picture of a typical local server used in the CultureBee system. Software is running on the local server to save the collected data in the MySQL database.

Figure 21. Typical local server in CultureBee system.

3.6 Remote monitoring and control Remote monitoring and control functions are performed at the main server. The access can be done anywhere where Internet connection is available via the homepage www.culturebee.se. The main server is a computer server with a main database where all the local servers synchronize their local database to it. The main server is installed

Wireless sensor data transmission 41 ______with web service and all the tasks that can be issued by the local server can also be issue via the main server. In this way, the remote user can have full control over the system as he/she is at the local building where the WSN is deployed. Moreover, the main server has additional functions as real-time graphical display or download of the collected data within a selectable time frame.

3.6.1 Demonstration board for remote control The remote control function in the CultureBee system proposed in Paper VII is implemented as a demonstration board. Note that the demonstration board is for controlling light intensity, but the same principle can be used for controlling other functions such as temperature and relative humidity. Figure 22 shows a block diagram and a photo of the demonstration board, which contains two dimmable LED lights on the top, an ambient light sensor and a dim-controller at the bottom from the left to the right. To each dimmable LED light, there is a ZigBee router device connected to it. The router device can be used both as a router and as an end node. The ambient light sensor and the dim controller are implemented as an end-device. The dim controller can dim the light from 10 to 100 % light intensity and the ambient light sensor gives feedback about the light intensity in the surrounding area to the dimmable lights so a close-loop light intensity output can be set. All the devices on the demonstration board can directly communicate with the local server to report the current status. When the ambient light sensor and the dim controller are moved out of range of the local server, they can re-route the signals via the routers as self-repair and forward the signal to the local server.

42 Remote monitoring and control ______

a) block diagram

b) photo Figure 22. Block diagram and photo of the demonstration board.

The local server is synchronized with the CultureBee homepage for presenting of the dimmable light status and to pull any control signal sent from the homepage. Figure 23 shows the control windows at the CultureBee homepage, similar to the windows that can be found at the local server, where one can turn the dimmable lamps on/off. The dimmable function is also implemented to set the lamps to a predefined light intensity with and without the feedback from the ambient light sensor. To perform a remote control, the Internet connection between the local server and the main server must be available at the time when the remote control command is set.

Wireless sensor data transmission 43 ______

Figure 23. Remote control window at the CultureBee homepage.

3.7 Future work The CultureBee system is currently running stably with five different WSN deployed at different buildings and locations. Although the developed CultureBee System has been functioning well, it leaves some improvement work to be done, e.g., further reducing the current consumption to prolong the battery lifetime. As a design for cultural buildings, a fully wireless solution would have been a better solution. In the ZigBee network, the router needs to be awake all the time to listen if any communication is going on. This means that a constant current supply is needed, which is not always available in an old cultural build. In such a case, WirelessHART and ISA SP100 would be a better solution. However, those solutions will result in higher cost. One can also consider powering the ZigBee routers with autonomous solutions. Autonomous wirelesses devices are self-powered by locally harvested energy, e.g., from ambient heat, light, electromagnetic waves, vibrations or air flow. The harvested power should be larger than the power consumption by the router to minimize the maintenance. By doing this it will gain the flexibility of the placement of the ZigBee router and deployment of the ZigBee sensor network. As shown in Table 6, the Sensing Data mode for ITN ZigBee end-device is the operation mode with highest duty cycle and thereby also the most current consuming mode. With the implementation used today, after the microcontroller has issued the sensor device for sensing data, CC2430 will be waiting for reply from the sensor. While waiting for the reply, CC2430 is unnecessary draining precious current from the battery. The author suggests that the Sensing Data mode could be improved as the flow diagram illustrated in Figure 24. The CC2430 radio chip should enter into the deep

44 Remote monitoring and control ______sleep-mode after issued the sensor for sensing data. When the sensor is done sensing data, it should issue an interrupt to CC2430 to wake it up from the sleep-mode and process the data.

Figure 24. Flow diagram of sensing data mode to save current consumption.

Deployed WSN in buildings with thick walls made of stones have shown to be troublesome to establish a network to fully cover the entire building. Switching to channel 1, which is locating at 868 MHz, might be a better solution. This is due to the fact that the signal path loss is less at a lower carrier frequency. As shown by the Friis formula (53), lowering the frequency to channel 1 can ideally reduce the path loss by 8.7 dB. Another benefit at 868 MHz is interaction with biological objects is less sensitive, which will improve the radio link in environments with many obstacles.

Wireless sensor data transmission 45 ______

However, lowering the frequency also gives larger physical size of the antenna solution. Roughly at one third of the carrier frequency used today (2.4 GHz), the antenna size will be roughly three times larger.

Results 47 ______

4 Results

This dissertation has dealt with both high-speed wired data transmission and wireless data transmission for wireless sensor networking. The work started with analyzing wired parallel-data transmission on a flexible foil material by simulations, prototyping and measurements. Parameters to study were rise/fall time, skin effect, conductor surface roughness and dielectric loss. A flex-rigid cable concept has been proposed as an alternative parallel-cable using the zebra connector to mate the cable with the connecting board. Effects from the via-holes and the connector has been simulated and measured to show that the size of the via-pads and the via-holes can affect the characteristic impedance considerably. It has also been shown that the connector can be co-designed in combination with the connecting pads so discontinuity of the characteristic impedance can be minimized. Data transmission rate at 2 Gbps was achieved with a 490-mm-long microstrip on the flexible cable with crosstalk taken into account. Parallel data transmission has been studied and the author noticed that there is a weakness in the commonly used mixed-mode S-parameter conversion method. The commonly used conversion method works well when couplings between parallel transmission lines are weak or do not exist, which it is not true in most cases. When designing differential transmission lines, it is of interest to strive for a high coupling between the conductors. A modification of the conversion method was proposed and simulations were performed to show that the result from the proposed conversion is more consistent with the simulation of coupled parallel transmission lines in comparison with the commonly used mixed-mode conversion technique. A design methodology for a fully differential RF and microwave front-ends was presented. Conversion between balanced and unbalanced signals was avoided to

48 Results ______maintain differential signaling for high noise immunity. Baluns are omitted to avoid narrowing the bandwidth of the system and unnecessary signal loss. Differential designs of the RF filter and matching network have been presented in order to verify the correctness of the design methodology. A platform of wireless remote monitoring and control system named CultureBee has been established for various deployments at different cultural buildings. So far, five networks have been deployed for data collection. The result will be used for analysis in order to get a better understanding of the indoor environment for preserving cultural heritage. The platform for wireless remote monitoring and control system is designed with a modular architecture to ease future improvement and extension of functions. In addition to the standard ZigBee protocol, the routers in the mesh network have been enriched with extra power amplifiers and low-noise amplifiers in order to extend the radio range at the cost of increased power consumption. Both the outdoor and indoor radio range performances have been conducted to get a better understanding of the behavior of the ZigBee-based wireless sensor network. Power consumption of the modules has been measured and the total battery lifetime has been calculated. Optimizations of both the hardware and software have been done in order to achieve long battery lifetime, low latency and robust networking. The developed remote monitoring function has been proven to work stably for almost one year now. The remote control functions have been implemented as a demonstration board integrated into the CultureBee platform as a proof of concept. The CultureBee platform has received a nomination for the Swedish Embedded Award 2010 and demonstrated at the Scandinavia Embedded Conference 2010 in Stockholm, Sweden.

Summary of the included papers 51 ______

5 Summary of the Included Papers

Paper 1 Study of High-Speed Data Transfer Utilizing Flexible and Parallel Transmission Lines by Allan Huynh, Shaofang Gong and Leif Odselius, Proceedings of the International Microelectronics And Packaging Society Nordic conference, Törnsberg, Norway, pp. 230-234, September 2005.

In this work, simulations on high-speed single-ended and parallel channels utilizing microstrips and striplines have been done to show parameters that limit the channel bandwidth and thus the data rate. It is shown that both conductor surface roughness and dielectric loss introduce extra AC noise to the signal. When having parallel conductors the AC noise increases with decreased rise/fall time, introduced by the skin effect, conductor surface roughness and dielectric loss.

Author’s contribution: The author set-up and performed the simulations of the design and wrote the manuscript. Deployed the networks at different places of interest and wrote the article.

Paper 2 High-Speed Parallel Data Transmission Utilizing a Flex-Rigid Concept by Allan Huynh, Pär Håkansson, Shaofang Gong and Leif Odselius, Proceedings of GigaHertz 2005, Uppsala, Sweden, pp. 206-209, November 2005.

In this work, transmission lines utilizing microstrips on a flexible foil with low dielectric loss (tanδ = 0.002) have firstly been simulated to compare with the

52 Summary of the included papers ______transmission lines laminated with a rigid part, a so-called Flex-rigid structure. Prototypes are then made for measurements. It is shown that the inductive and capacitive effects from the via-hole and via-pads on the Flex-rigid structure can be designed such that they compensate each other, resulting in a smooth transition of the characteristic impedance. This gives the advantage when using the structure for high- speed designs.

Author’s contribution: The author proposed the solution, set-up and performed the simulations, designed and measured the structure and wrote the manuscript.

Paper 3 High-Speed Board-To-Board Interconnects Utilizing Flexibel Foils and Elastomeric Connectors by Allan Huynh, Pär Håkansson, Shaofang Gong and Leif Odselius, Proceedings of The 8th IEEE CPMT International Conference on High Density Microsystem Design, Packaging and Component Failure Analysis, Shanghai, China, pp. 139-142, June 2006.

This work presents a board-to-board interconnect technique utilizing elastomeric connectors and parallel microstrip lines on a flexible foil cable with low dielectric loss (tanδ = 0.002). It is shown that a pad structure combined with an elastomeric connector can be co-designed such that a good signal integrity and thus a high data transmission rate is achieved. It is also shown that a 2 Gbps data transmission rate can be achieved with a 490-mm-long microstrip on the flexible cable, where crosstalk is taken into account.

Author’s contribution: The author proposed the solution, set-up and performed the simulations, designed and measured the cables and wrote the manuscript.

Paper 4 Single-ended to Mixed-Mode S-Parameter Conversion for Networks with Coupled Differential Signaling by Allan Huynh, Pär Håkansson and Shaofang Gong, Proceedings for 36th European Microwave Conference 2007, Munich, Germany, pp. 238-241, October 2007.

Summary of the included papers 53 ______

In this work, it is shown that the commonly used method for converting from standard single-ended to mixed-mode S-parameters for networks with differential signaling gives an insufficient accuracy for coupled differential transmission lines. A correct conversion matrix equation must include the odd- and even-mode impedances which are not equal to the unique characteristic impedance owing to signal coupling in the network. Otherwise, the conversion is only valid for uncoupled differential signals.

Author’s contribution: The author observed the weakness in mixed-mode conversion method commonly used and proposed an improved solution. The author did most of the calculations to find the new conversion method, set-up and simulated a differential microstrip for verification of the new conversion method and wrote most of the manuscript.

Paper 5 Truly Differential RF and Microwave Front-End Design by Shaofang Gong, Allan Huynh, Magnus Karlsson, Adriana Serban, Owais and Joakim Östh, Proceedings of IEEE Wireless and Microwave Technology Conference WAMICON 2010, Florida, USA, pp. 1-5, April 2010.

In this work, a new design methodology for truly differential RF and microwave front-ends has been presented. Baluns are avoided using this design methodology, while achieving differential signaling for high noise immunity. A case study on an ultra-wide band RF front-end in the frequency band 6-9 GHz has been performed using the new design methodology, indicating that both wide bandwidth and high performance can be achieved using this design methodology. A direct comparison between single-ended and differential designs of the RF filter has also been presented in order to verify the correctness of the design methodology.

Author’s contribution: The author was involved in the concept study and discussions and wrote part of the manuscript.

54 Summary of the included papers ______

Paper 6 Wireless Remote System Monitoring for Cultural Heritage by Allan Huynh, Jingcheng Zhang, Qin-Zhong Ye and Shaofang Gong, Sensors & Transducers Journal (ISSN 1726-5479), Vol. 118, Issue 7, pp. 1-12, July 2010.

In this work, a wireless remote sensor network based on the ZigBee technology together with a simplified data collection system is presented. The wireless sensor devices can automatically join available network or when deploying a new one. The collected data is automatically and periodically synchronized to a remote main server via an Internet connection. The data can be used for centralized monitoring and other purpose. The power consumption of the sensor module is minimized and the battery lifetime is estimated up to 10 years.

Author’s contributions: Design of the system architecture and all the hardware needed for the wireless sensor network. Wrote the article.

Paper 7 Design of the Remote Climate Control System for Cultural Buildings Utilizing ZigBee Technology by Jingcheng Zhang, Allan Huynh, Qin-Zhong Ye and Shaofang Gong, Sensors & Transducers Journal (ISSN 1726-5479, Vol. 118, Issue 7, pp. 13-27, July 2010.

In this work, a wireless solution of remote climate control for cultural buildings is presented. The system allows users to use web service to control climate in different cultural buildings, like churches. The wireless sensor networks deployed in churches receive the control commands and manage the indoor climate. The whole system is modularly designed, which makes possible an easy service extension, system reconfiguration and modification. This paper includes the system overview and the software design of each part within the system.

Author’s contribution: Hardware design, discussions and proposed some of the technical software implementations in the system.

Summary of the included papers 55 ______

Paper 8 ZigBee Radio with External Low-Noise Amplifier by Allan Huynh, Jingcheng Zhang, Qin-Zhong Ye and Shaofang Gong, Sensors & Transducers Journal (ISSN 1726-5479), Vol. 114, Issue 3, pp. 184-191, March 2010.

In this work, a performance study of a ZigBee radio module with an external low-noise amplifier is presented with measurements in both outdoor and indoor environments. Previous study with standard ZigBee radio has already shown that the indoor campus environment such as walls and floors would reduce the radio link range drastically and the packet error rate increased. In this study, an external low-noise amplifier has been added to a ZigBee radio module to increase the receiver sensitivity. By increasing the receiver sensitivity, the radio range can be increased without increasing of the radio power output and the power consumption can still be kept low to obtain long battery lifetime.

Author’s contribution: Design of the ZigBee radio module with an external low- noise amplifier and verify that the radio is functional both in the lab and in the deployment field. Performed measurements about the power consumption and indoor radio range in the campus environment and wrote the article.

Paper 9 ZigBee radio with External Power Amplifier and Low-Noise Amplifier by Allan Huynh, Jingcheng Zhang, Qin-Zhong Ye and Shaofang Gong, Sensors & Transducers Journal (ISSN 1726-5479), Vol. 118, Issue 7, pp. 110-121, July 2010.

In this work, the performance study of a ZigBee radio module with both an external power amplifier and a low-noise amplifier, measured in outdoor and indoor environments, respectively. Both the external power amplifier and the low-noise amplifier have been added to a ZigBee module to increase both the transmitter power and receiver sensitivity. The power consumption issue with the added amplifiers is studied as well, indicating that the module can still be battery driven with a battery lifetime of about 9 years at a normal sampling rate of the sensor.

56 Summary of the included papers ______

Author’s contribution: Design of the ZigBee radio module with an external power amplifier and a low-noise amplifier. Verify that the radio is functional both in the lab and in the deployment field. Performed the power consumption and radio range measurements and wrote the article.

Paper 10 Reliability and Latency Enhancements in a ZigBee Remote Sensing System by Jingcheng Zhang, Allan Huynh, Qin-Zhong Ye and Shaofang Gong, Proceedings of The Fourth International Conference on Sensor Technologies and Applications (SENSORCOMM 2010), Venice/Mestre, Italy, pp. 196-202, July 2010.

In this work, methods to improve the reliability and optimize the system latency of our own-developed ZigBee remote sensing system are introduced. The concept of this system utilizes the ZigBee network to transmit sensor information and process them at both local and remote databases. The enhancement has been done in different parts in this system. In the ZigBee network part, the network topology is configured and controlled. The latency for message transmitting is also optimized. In the data processing part, the network status check function and data buffer function are introduced to improve the system reliability. Additionally, the system latency is measured to compare it with that from the Ad-hoc on demand distance vector algorithm used in the ZigBee standard.

Author’s contribution: Hardware design and set-up the measurement equipment. Proposed some of the technical software solution in the system.

Bibliography 57 ______

BIBLIOGRAPHY [1] Reinhold Ludwig and Pavel Bretchko, RF Circuit Design Theory and Applications, Prectice Hall, USA, 2000, pp. 37-86.

[2] David M. Pozar, Microwave Engineering Third Edition, John Wiley & Sons Inc., USA, 2005, pp 174-183, 341-345, 493-500 and 647-649.

[3] Stephen H. Hall, Garrett W. Hall, James A. McCall, High-Speed digital System Design, John Wiley & Sons, Inc., USA, 2000, pp. 9-19.

[4] Wolfgang Peter Siebert, “High frequency cable connector for twisted pair cables” IEEE transactions on components and packaging technologies, Vol 26, No 3, pp. 642-650, Sep 2003.

[5] Allan Huynh, Pär Håkansson, Shaofang Gong and Leif Odselius, “High-Speed Board-To-Board Interconnects Utilizing Flexibel Foils and Elastomeric Connectors”, Proceedings of The 8th IEEE CPMT International Conference on High Density Microsystem Design, Packaging and Component Failure Analysis, Shanghai, China, pp. 139-142, June 2006.

[6] Allan Huynh, Shaofang Gong and Leif Odselius, “Study of High-Speed Data Transfer Utilizing Flexible and Parallel Transmission Lines”, Proceedings of The International Microelectronics And Packaging Society Nordic, Törnsberg, Norway, pp. 230-234, September 2005.

[7] Allan Huynh, Pär Håkansson, Shaofang Gong and Leif Odselius, “High-Speed Parallel Data Transmission Utilizing a Flex-Rigid Concept”, Proceedings of GigaHertz 2005, Uppsala, Sweden, pp. 206-209, November 2005.

[8] Shaofang Gong, Hjalmar Hesselbom, Christer Svensson, Thomas Lewin, “Simulation of coupled striplines for differential data transmission” report, IMC 1994, Linköping, Sweden, 1994.

58 Bibliography ______

[9] National LVDS Group, LVDS Owners’s Manual – 3rd Edition, National Semiconductor LVDS.national.com, Spring 2004.

[10] Ning Yang, Christophe Caloz and Ke Wu, “Greater Than the Sum of Its Parts” IEEE Microwave Magazine, pp. 69-82, June 2010.

[11] K. Kurokawa “Power Waves and the Scattering Matrix” IEEE Transactions of Microwave Theory and Techniques, pp. 194-202, Mars 1965.

[12] David E. Bockelman, William R. Eisenstadt, “Combined Differential and Common-Mode Scattering Parameters: Theory and Simulation” IEEE transactions on microwave theory and techniques, Vol. 43, No. 7, pp.1530- 1539, July 1995.

[13] William R. Eisenstdt, Bob Stengel and Bruce M. Thompson, Microwave Differentail Circuit Design, Artech House Inc., Norwood, MA, USA 2006.

[14] MAXIM High-Frequency/Fiber Communications Group “Single-Ended and Differential S-Parameters” Application Note: HFAN-5.1.0 Rev0; 03/01, November 2005.

[15] Anritsu Microwave Measurements Division “Three and Four Port S-parameter Measurements” Application Note, Anritsu, May 2002.

[16] J. Martens, D. Judge and J. Bigelow, “Multiport Vector Network Analyzer Measurements” IEEE Microwave Magazine, pp. 72-8, December 2005.

[17] Ken Vaz and Michael Caggiano, “Measurement Technique for the Extraction of Differential S-Parameters from Single-ended S-Parameters” Proceedings of IEEE 27th International Spring Seminar on Electronics Technology, pp. 313- 317, 2004.

[18] Allan Huynh, Pär Håkansson and Shaofang Gong, “Single-ended to Mixed- Mode S-Parameter Conversion for Networks with Coupled Differential

Bibliography 59 ______

Signaling”, Proceedings of 36th European Microwave Conference 2007, Munich, Germany, pp. 238-241, October 2007.

[19] Vern A. Dubendorf, Wireless Data Technologies, John & Wiley, Great Britain, 2003, pp. 1-6.

[20] Jochen Schiller, Mobile Communications – Second Edition, Addison Wesley, Great Britain, 2003, pp. 7–15.

[21] Chee-Yee Chong and Srikanta P. Kumar, “Sensor networks: Evolution, opportunities, and challenges”, Proceedings of IEEE, Vol. 91, no. 8, pp. 1247– 256, August 2003.

[22] Xia-Ming, Doug Yabo, Lu Dongming, “A Wireless Sensor System for Long- Term Microclimate Monitoring in Wildland Cultural Heritage Sites”, Proceedings of International Symposium on Parallel and Distributed Processing with Applications 2008, Sydney, NSW, pp 207-214. December 2008.

[23] Växjö Stift, Klimatmätning – en studie av tekniklösningar och framtida utvecklingsmöjligheter, ’Project report for Växjö dioces (in Swedish)’, Arkitekt & Miljö AB, Växjö, Sweden, January 2010.

[24] http://www.culturebee.se (online October 2010)

[25] http://www.ZigBee.org (online October 2010)

[26] http://standards.ieee.org/getieee802/download/802.15.4-2006.pdf

[27] http://www.hartcomm.org (online October 2010)

[28] http://www.isa.org (online October 2010)

[29] Atmel Corporation, T7024, Datasheet. (online October 2010)

60 Bibliography ______

[30] William Stallings, Wireless Communications and Networks, Prectice Hall, USA, 2002, pp. 110-113.

[31] Tomas H. Lee, The Design of CMOS Radio-Integrated Circuits, New York: Cambridge Press, 1998, pp. 77–80.

[32] http://en.wikipedia.org/wiki/Fresnel_equations (online October 2010)

[33] David Halliday, Robert Resnick and Jearl Walker, Fundamentals of Physics 7th Edition, John Wiley & Sons, Inc, USA, 2005, pp. 904-915.

[34] Behzad Razavi, RF Microelectronics, Prentice Hall, USA, 1998, pp. 37-49.

[35] David M. Pozar, Microwave and RF Design of Wireless Systems, John Wiley & sons inc., USA, 2000, pp. 87-93.

[36] M. Loy, “Understanding and Enhancing Sensitivity in Receivers for Wireless Applications” Application note Texas Instrument Inc, 1999-05.

[37] Allan Huynh, Jingcheng Zhang, Qin-Zhong Ye and Shaofang Gong ”ZigBee radio with External Low-Noise Amplifier”, Sensors & Transducers Journal (ISSN 1726-5479), Vol. 114, Issue 3, pp. 184-191. March 2010.

[38] Hawar Ramazanali, Jonas Olsson, Johan Lönn, Allan Huynh, Qin-Zhong Ye and Shaofang Gong, “Evaluation of ZigBee Networking in a Campus Enviroment” ISAST Transactions on Communications and Networking, pp. 30- 35, 2008.

[39] Allan Huynh, Jingcheng Zhang, Qin-Zhong Ye and Shaofang Gong, ”ZigBee radio with External Power Amplifier and Low-Noise Amplifier” Sensors & Transducers Journal (ISSN 1726-5479), Vol. 118, Issue 7, pp. 110-121, July 2010.

Bibliography 61 ______

[40] V. Raghunathan , C. Schurgers , S. Park , and M. B. Srivastava, “Energy - aware wireless microsensor networks”, IEEE Signal Processing Magazine, vol. 19 , no. 2 , Mars 2002 , pp. 40 – 50.

[41] Jingcheng Zhang, Allan Huynh, Qin-Zhong Ye and Shaofang Gong, “Remote Sensing System for Cultural Buildings Utilizing ZigBee Technology” Proceedings of the 8th International Conference on Computing, Communications and Control Technologies (CCCT 2010), Orlando, USA, pp. 71-77, April 2010.

[42] J.M. Rabaey, Digital Integrated Circuits: A Design Perspective, Prentice Hall, USA, 1996, pp. 209-211.

[43] Hynek Raisigel, Gilles Chabanis, Isabelle Ressejac and Michel Trouillon, “Autonomous Wireless Sensor Node for Building Climate Conditioning Application” Proceedings of the 4th International Conference on Sensor Technologies and Applications” Venice, Italy, pp. 68-73, July 2010.