And Mission-Critical Applications in Industrial Wireless Sensor Networks

Enabling Time- and Mission-Critical Applications in Industrial Wireless Sensor Networks

Hossam Farag

Department of Information Systems and Technology Mid Sweden University Licentiate Thesis No. 151 Sundsvall, Sweden 2019 Mittuniversitetet Informationssystem och -teknologi ISBN 978-91-88527-84-4 SE-851 70 Sundsvall ISNN 1652-8948 SWEDEN Akademisk avhandling som med tillstand˚ av Mittuniversitetet i Sundsvall framlagges¨ till offentlig granskning for¨ avllaggande¨ av teknologie licentiatexamen Onsdagen den 30 januari 2019 i M102, Mittuniversitetet, Holmgatan 10, Sundsvall. c Hossam Farag, 2019 Tryck: Tryckeriet Mittuniversitetet My Wife My Parents iv Abstract

Nowadays, Wireless Sensor Networks (WSNs) ”have gained importance as a flexible, easier deployment/maintenance and cost-effective alternative to wired networks, e.g., Fieldbus and Wired-HART, in a wide-range of applications. Initially, WSNs were mostly designed for military and environmental monitoring applications where energy efficiency is the main design goal. The nodes in the network were expected to have a long lifetime with minimum maintenance while providing best-effort data delivery which is acceptable in such scenarios. With recent advances in the industrial domain, WSNs have been subsequently extended to support industrial automation applications such as process automation and control scenarios. However, these emerging applications are characterized by stringent requirements regarding reliability and real-time communications that impose challenges in the design of Industrial Wireless Sensor Networks (IWSNs) to effectively support time- and mission-critical applications. Typically, time- and mission-critical applications support different traffic categories ranging from relaxed requirements, such as monitoring traffic to firm requirements, such as critical safety and emergency traffic. The critical traffic is mostly acyclic in nature and occasionally occurs at unpredictable time instants. Once it is generated, it must be delivered within strict deadlines. Exceeding the delay bound could lead to system instability, economic loss, or even endanger human life in the working area. The situation becomes even more challenging when an emergency event triggers multiple sensor nodes to transmit critical traffic to the controller simultaneously. The unpredictability of the arrival of such a type of traffic introduces difficulties with regard to making a suitable scheduling that guarantees data delivery within deadline bounds. Existing industrial standards and related research work have thus far not presented a satisfactory solution to the issue. Therefore, providing deterministic and timely delivery for critical traffic and its prioritization over regular traffic is a vital research topic. Motivated by the aforementioned challenges, this work aims to enable real-time communication for time- and mission-critical applications in IWSNs. In this context, improved Medium Access Control (MAC) protocols are proposed to enable a priority-based channel access that provides a timely delivery for acyclic critical traffic. The proposed framework starts with a stochastic modelling of the network delay performance under a priority-oriented transmission scheme, followed by two MAC approaches. The first approach proposes a random Clear Channel Assess-

v vi

ment (CCA) mechanism to improve the transmission efficiency of acyclic control traffic that is generated occasionally as a result of observations of an established tendency, such as closed-loop supervisory traffic. A Discrete-Time Markov Chain (DTMC) model is provided to evaluate the performance of the proposed protocol analytically in terms of the expected delay and throughput. Numerical results show that the proposed random CCA mechanism improves the shared slots approach in WirelessHART in terms of delay and throughput along with better transmission reliability. The second approach introduces a slot-stealing MAC protocol based on a dynamic deadline-aware scheduling to provide deterministic channel access in emergency and event-based situations, where multiple sensor nodes are triggered simultaneously to transmit time-critical data to the controller. The proposed protocol is evaluated mathematically to provide the worst-case delay bound for the time-critical traffic and the numerical results show that the proposed approach out- performs TDMA-based WSNs in terms of delay and channel utilization. Acknowledgements

Praise be to Allah I would like to take this opportunity to express my heartfelt appreciation to the following persons who have contribute directly or indirectly to the completion of this work. Firstly, my sincere gratitude to my supervisor, Prof. Mikael Gidlund, for his constant support and guidance in my research work. I would like to thank him for his continuous support, insightful suggestions, encouraging feedback, and the freedom in choosing research directions. Thanks to my co-supervisor Dr. Patrik Osterberg¨ for his support, useful com- ments and constructive criticisms that have helped to improve this work. Also, I would like to thank him for his administrative support in other department-related issues. Thanks to Dr. Aamir Mahmood for his excellent cooperation and discussions during my research work. Thanks to Simone Grimaldi, Teklay Gebremichael, Raul` Rondon,` Mehrzad Lavassani and all other colleagues at the Department of Informa- tion Systems and Technology, Mid Sweden University for their kindness and friend- liness. Last but not least, special thanks to my dear wife, Samira, for her endless love, support and for standing behind me in all what I do. Thanks to all my family and friends for supporting me spiritually throughout my life.

vii viii Contents

Abstract v

Acknowledgements vii

List of Papers xi

Terminology xv

1 Introduction 1 1.1 Overview ...... 1 1.2 Problem Statement ...... 2 1.3 Overall Aim and Research Topic ...... 4 1.4 Methodology ...... 4 1.5 Contributions ...... 6 1.6 Thesis Outline ...... 6

2 Background and Related Work 7 2.1 Overview of IWSNs ...... 7 2.2 Overview of the Industrial Standards for PA Applications ...... 8 2.2.1 WirelessHART ...... 8 2.2.2 ISA100.11a ...... 9 2.2.3 WIA-PA ...... 9 2.2.4 Zigbee ...... 10 2.2.5 IEEE 802.15.4e ...... 10 2.3 MAC Protocols in IWSNs ...... 10 2.3.1 General Taxonomy of Wireless MAC Protocols ...... 11

ix x CONTENTS

2.3.2 MAC Protocols in IWSNs Standards ...... 12 2.3.3 Related Work ...... 15

3 Priority-Based Real-Time Communication in IWSNs 19 3.1 Overview ...... 19 3.2 Modelling Priority-Oriented Packet Transmissions ...... 19 3.3 Improving the Transmission Efﬁciency of Acyclic Trafﬁc in IWSNs . . 21 3.4 Deterministic Real-Time Communication of Multiple Critical Flows . . 22

4 Summary of Publications 27

5 Conclusions and Future Work 33 5.1 Concluding Remarks ...... 33 5.2 Ethical and Societal Considerations ...... 34 5.3 Future Work ...... 35

Bibliography 37

Paper I 43

Paper II 59

Paper III 75

Biography 99 List of Papers

This thesis is mainly based on the following papers, herein referred by their Roman numerals:

Paper I Priority-Oriented Packet Transmissions in Internet of Things: Modeling and Delay Analysis Lakshmikanth Guntupalli, Hossam Farag, Aamir Mahmood, and Mikael Gidlund In Proceedings of IEEE International Conference on Communications (ICC 2018), Kansas City, USA, May 2018.

Paper II PR-CCA MAC: A Prioritized Random CCA MAC Protocol for Mission- Critical IoT Applications Hossam Farag, Aamir Mahmood, Mikael Gidlund, and Patrik Osterberg¨ In Proceedings of IEEE International Conference on Communications (ICC 2018), Kansas City, USA, May 2018.

Paper III A Delay-Bounded MAC Protocol for Mission- and Time-Critical Applications in Industrial Wireless Sensor Networks Hossam Farag, Mikael Gidlund, and Patrik Osterberg¨ In IEEE Sensors Journal, Vol. 18, No. 6, pp. 2607-2616, March 2018.

xi xii List of Figures

1.1 Example of future industrial automation networks [25]...... 2 1.2 Trafﬁc categories in PA applications and their corresponding latency requirements...... 3 1.3 Research work ﬂow...... 5

2.1 Protocol stack comparison of industrial standards for PA applications.9 2.2 Wireless MAC protocols taxonomy...... 11 2.3 The superframe structure of Zigbee, WIA-PA, WirelessHART and ISA100.11a [9]...... 13 2.4 TSCH superframe...... 14 2.5 DSME multi-superframe structure...... 14

3.1 Performance comparisons...... 21 3.2 Slot timing structure of the proposed PR-CCA protocol...... 21 3.3 Performance comparisons...... 22 3.4 SS-MAC channel access scenario...... 23 3.5 Worst-case delay comparison versus different number of nodes. . . . . 25 3.6 Utilization comparisons...... 25

xiii xiv Terminology

Abbreviations and Acronyms

CAN Controller Area Network CAP Contention Access Period CCA Clear Channel Assessment CDMA Code Division Multiple Access CFP Contention Free Period CSMA/CA Carrier Sense Multiple Access with Collision Avoidance CTS Clear-To-Send DSME Deterministic and Synchronous Multi-channel Extension DTMC Discrete-Time Markov Chain FA Factory Automation FDMA Frequency Division Multiple Access GTS Guaranteed Time Slot HART Highway Addressable Remote Transducer HCF HART Communication Foundation ISA International Society of Automation IWSN Industrial Wireless Sensor Network LLDN Low-Latency Deterministic Network MAC Medium Access Control MEMS Micro Electro-Mechanical Systems PA Process Automation RTS Request-To-Send TDMA Time Division Multiple Access TSCH Time Slotted Channel Hopping TSMP Time Synchronized Mesh Protocol WIA-PA Wireless networks for Industrial Automation-Process Automa- tion WSN Wireless Sensor Network

xv xvi Chapter 1

Introduction

1.1 Overview

With recent advances in wireless communications and Micro Electro-Mechanical Systems (MEMS) technology, Wireless Sensor Networks (WSNs) found their way into a wide range of applications and systems where sensor nodes harvest information from their physical environment. Initially, WSNs were mostly designed for military, environmental monitoring and healthcare applications [1], where operation with little or no human intervention for long periods of time is required. Therefore, energy efficiency was the main design goal for such applications, as the nodes in the network were expected to have long lifetime. The field of industrial automation has introduced effective and low-cost solutions in order to improve the productivity and efficiency of the industrial process[2]. In a typical field network such as the one shown in Fig. 1.1, sensor nodes are distributed in the working field to cap- ture real-world measurements, such as temperature, pressure and vibration, and for- ward the collected information to a central node. Based on the received information, the controller may accordingly send control commands to the actuators to perform necessary control actions. Traditionally, industrial automation systems are realized through wired communications, which are served by fieldbus systems [3] such as PROFIBUS [4], wired-Highway Addressable Remote Transducer (wired-HART) [5] or Controller Area Network (CAN) [6]. The concept of Industrial Wireless Sensor Networks (IWSNs) is introduced as an alternative to the wired solution to enable low-cost and low-power wireless technology in the industrial domain, and hence offers competitive advantages. First of all, the cost and time needed for the installation and maintenance of the large number of cables normally required in the industrial environment can be substantially reduced, thus making plant set-up and reconfiguration easier, especially in such harsh environments where chemicals, vibrations, or moving parts exist and could potentially damage any sort of cabling. The industrial market [7] estimates that IWSNs enable cost savings of up to 90% compared to the deployment cost of wired field devices in

1 2 Introduction

Figure 1.1: Example of future industrial automation networks [25]. the industrial automation domain. Furthermore, IWSNs provide great availability as it can offer built-in redundancy and capabilities for anticipatory system maintenance and failure recovery. Flexibility of adding new nodes to the network and operate in different network topologies is another added advantage. However, in order to satisfy the strict real-time requirements of time- and mission- critical applications in the industrial domain, the design of IWSNs introduces many challenges. In this context, real-time means that the system can only function properly, and sometimes safely, if data arrives in a timely and reliable fashion. Unaccept- able delay in conveying sensed information can result in undesired complications such as, damage to the equipment or threat to human life [8]. For instance, in a plastic extrusion plant, reporting excessive melt pressure in a timely fashion is crucial in order to avoid explosion. Moreover, the critical trafﬁc in these applications is mostly acyclic in nature which, imposes a challenge to schedule such trafﬁc to meet its corresponding time constraints. This research work mainly focuses on how to support real-time communication for such applications.

1.2 Problem Statement

Most existing network protocols are designed primarily to provide energy efﬁ- cient performance for a longer lifetime for the network while providing only best- effort performance for data delivery. Such behaviour is adequate for applications where the communication delay can be tolerated, i.e., collected data is not used to trigger actions immediately. However, this best-effort performance is no longer suitable for time- and mission-critical applications, such as ﬁre alarms, leakage of poi- 1.2 Problem Statement 3

Figure 1.2: Trafﬁc categories in PA applications and their corresponding latency requirements.

sonous gases, motion adaptation for conveyor belt movements, affiliated robotics and temperature control of a furnace. These applications are characterized by stringent real-time requirements, that is the system can only function properly, and sometimes safely, if data arrives in a timely and reliable fashion [9]. The International Society of Automation (ISA) committee defines the traffic within the industrial Process Automation (PA) domain into safety, control and monitoring [10] as shown in Fig. 1.2, in which requirements on latency and reliability vary. The available IWSNs standards for PA applications are Zigbee [11], Wire- lessHART [12,13], ISA 100.11a [12], and Wireless network for Industrial Automation- Process Automation (WIA-PA) [14], which are all based on the IEEE 802.15.4 standard. Among all these standards, WirelessHART is considered as the major and dominant standard already introduced on the market to bring wireless technology to the process field. Recently, the IEEE 802.15.4e [15, 16] emerged to overcome 802.15.4 MAC protocol limitations. The goal was to define a low-power multi-hop MAC protocol, capable of addressing the emerging needs of industrial PA applications. The aforementioned standards manage to serve Class 5 and Class 4 of the monitoring traffic as well as the non-critical control traffic of Class 3 and Class 2. However, none of them defines how to efficiently manage the transmission/retransmission of the event-based critical traffic of Class 1 and Class 0 in time- and mission-critical applications. This critical traffic requires reliable delivery to the sink node within stringent deadlines. Late delivery of this traffic or disruption of its ongoing communication is considered a critical failure. Guaranteeing deterministic delay for this traffic and its prioritization over other regular traffic is a key challenge. Moreover, some application scenarios, such as fire emergency situations where a combination of sensors, e.g., smoke, temperature and CO are triggered to detect the presence of fire when its sensor reading is above a specified threshold, require the delivery of 4 Introduction

multiple triggered data having the same priority (importance). This introduces a challenge of how to schedule the transmissions to deliver each critical ﬂow within its corresponding deadline bound.

1.3 Overall Aim and Research Topic

The aim of this work is to push the existing IWSNs technology to efficiently support the event-based critical traffic of Class 0 and Class 1. In particular, the objective is to develop solutions to enable timely delivery for the non-deterministic critical traffic and its prioritization over the non-critical one along with efficient channel utilization in time- and mission-critical applications. Based on the discussion in the previous section, the research topic can be formulated into the following questions:

• RQ1: How can the effect of packet prioritization on the average delay performance under different trafﬁc conditions be modelled and analysed?

• RQ2: Acyclic trafﬁc is unpredictable in nature, how can timely delivery regarding the transmission/retransmission of such trafﬁc in IWSNs be enabled?

• RQ3: In emergency and safety-related situations in particular, multiple nodes are simultaneously triggered to send critical data to the controller within stringent deadline bounds. How can deterministic channel access for the triggered trafﬁc with a delay bounded performance be scheduled, while maintaining ef- ﬁcient channel utilization?

1.4 Methodology

The technical research steps in this work followed the approach illustrated in Fig. 1.3. This work started with a qualitative literature review aiming to identify the general research problem to be addressed throughout this work. The first cycle of the study was carried out with the goal of gaining the basic knowledge of IWSN technologies and standards, as well as the corresponding open issues and research challenges. At this point, providing real-time communications for time- and mission-critical industrial applications was identified as one of the most prominent challenges and was adopted as the research problem to be addressed in this work with more focusing on the delay performance. The next step was to define the approach to be taken in order to investigate the identified research problem in the previous step. It was decided to approach the research problem through the MAC layer by designing efficient MAC protocols to enable real-time communication for time- and mission-critical applications. This is because the MAC layer fundamentally determines the basic data transport capabilities of the network, and additional network mechanisms, such as routing or topology 1.4 Methodology 5

Figure 1.3: Research work ﬂow.

control, are often integrated into the MAC protocol or closely aligned with its design choices. Moreover, the MAC layer controls access to the physical shared medium which mainly controls the delay performance of the network. The following stage was to thoroughly review the recent work approaches that deal with the defined research problem, which also helped in identifying a number of research gaps. The identified research problem and the corresponding research gaps were then formulated into the research questions RQ1, RQ2 and RQ3. Next, a set of MAC protocols have been developed in order to investigate and approach the challenges highlighted by the aforementioned research questions. The basic idea of the proposed MAC protocols is to enable timely delivery for the unpredictable critical traffic by providing priority-based medium access schemes. RQ1 was initially addressed by introducing an analytical modelling approach for a priority-oriented packet transmission scheme. The aim was to develop analytical models to examine the average delay performance of priority-oriented transmissions over a duty-cycled MAC protocol in low as well as heavy traffic conditions. The next phase was to introduce two MAC protocols in order to attack the issues highlighted by RQ2 and RQ3. Probability theory and Markov chains are used to evaluate the proposed protocols analytically to provide an intuition on performance metrics such as the delay. Moreover, the effectiveness of the proposed protocols is revealed by introducing comparisons with the available industrial standards and state-of-the- art. The output of the aforementioned research work is presented in the contributed papers listed in the following section. Finally, a review was conducted to verify that the research objective has been reached based on the formulated research questions, and possible directions for future work were outlined. 6 Introduction

1.5 Contributions

The work in this thesis is summarized in the following papers, included in full at the end of this work:

• Paper I introduces a stochastic model for priority-oriented packet transmissions over a synchronous duty cycling MAC protocol. Two Discrete Time Markov Chain (DTMC) models are developed to examine the effect of packet prioritization on the average delay performance. A 2D DTMC model represents the priority and queuing behaviour at a node while a 1D DTMC models the active nodes in the network. Both models are fused together in order to evaluate the delay performance analytically. The accuracy of the models is demonstrated via discrete-event simulations as both results match precisely. The proposed work aims at addressing RQ1 and was also the kick-off for the remaining contributions. • Paper II aims to answer RQ2 by introducing a priority-based MAC protocol to handle the transmissions/retransmissions of acyclic critical data in a timely fashion. An efﬁcient random Clear Channel Assessment (CCA) mechanism is proposed to enable low collision probability between contending nodes which in turn improves the performance in terms of delay and throughput. • Paper III proposes SS-MAC, a slot stealing MAC protocol to provide deterministic real-time communication for time- and mission-critical applications in IWSNs. SS-MAC handles concurrent transmissions in emergency situations where multiple sensor nodes are simultaneously triggered to send critical data to the controller. The triggered nodes are deterministically scheduled by the controller node to transmit their critical data based on a dynamic deadline- aware schedule. The transmission of critical data is characterized by a deterministic delay bound to guarantee predictable performance. This work aims to answer RQ3.

1.6 Thesis Outline

The reminder of this thesis is organised as follows. Chapter 2 introduces a com- prehensive background and related work about the research topic. Chapter 3 de- scribes the proposed approach presented here. Contributed papers are summarized in Chapter 4 followed by the conclusion and suggested future work directions in Chapter 5. Chapter 2

Background and Related Work

2.1 Overview of IWSNs

During the past decade, WSNs have been introduced into a wide variety of applications [17], such as health monitoring of civil infrastructures [18], personal health monitoring [19], smart grid [20], agriculture [21], and industrial automation [10]. Adopting emerging wireless technology through the deployment of IWSNs offers competitive advantages over the wired solutions in the industrial domain, such as low-cost, flexibility, easy deployment/maintenance and self-configuration [10, 22]. This leads to an improvement of product quality, streamlining operations and speed- ing up production, which in turn revolutionizes the industrial processing and helps the industry meet the demands of increased competitiveness in the marketplace. The industrial domain can be divided into different domains, such as PA and Factory Automation (FA) [23]. In PA applications, the product is processed in a continuous manner (e.g. oil, gas, chemicals). In FA applications (e.g. automotive, med- ical, and the food industries) the products are processed in discrete steps, i.e., the products are assembled together using sub-assemblies or single components. FA is mainly characterized by short range communications (< 10 m) and its corresponding standards are mainly star networks, while PA has a longer range of communications (> 100 m), and the standards propose mesh networks. Moreover, since the discrete product in FA needs to be picked, assembled or palletized at high speeds, the sampling rate and real-time requirements are often stricter than those of PA. This work mainly focuses on IWSNs deployed in PA applications. The benefits that can be offered by IWSNs are facing certain bounds mainly due to the nature of the harsh environment in the industrial domain. These limitations lead to certain research challenges, including, but not limited to, the following:

• Real-time communication: Critical trafﬁc in PA applications, such as safety and closed loop regularity, are extremely delay sensitive and should be transmitted in s deterministic, timely and reliable fashion [24]. A too late arrival means lim-

7 8 Background and Related Work

ited use, and may cause system degradation, economical loss, or even worse, endanger human life [8]. • Service differentiation: Due to the inherent mixed-criticality property in a typical PA application, IWSNs should also be designed to provide service differentiation for wireless medium access. The challenge is how to provide the critical traffic with the highest priority to access the channel in conjunction with its unpredictable nature. • Coexistence and interference avoidance: A typical industrial site contains var- ious wireless networks and communication systems operating at the 2.4 GHz ISM band which creates interference of the radio signals [25]. • Safety: The automation equipment should be strictly designed to avoid or reduce the risk of uncontrolled or dangerous situations for the safety of humans, environment and property [23]. • Energy consumption: Nodes are required to be energy efficient to ensure a longer lifetime of the network and avoid frequent scheduled maintenance for thousands of nodes. This a serious problem for the industries that are expected to operate flawlessly around the clock [23]. • Security: The efficient integration of security mechanisms with the automation system and protecting the system against denial-of-service attacks is a key issue [26].

Based on the considered application requirements, some issues are more important than others. For instance, in time- and mission-critical applications, the deterministic real-time performance is a more important issue than the energy consumption [24]. The existing industrial standards for PA could not satisfy some issues in a satisfactory manner, and thus new research directions in this context are revealed. The work in this thesis mainly addresses the ﬁrst two requirements/challenges, which are closely related to time- and mission-critical applications.

2.2 Overview of the Industrial Standards for PA Ap- plications

The available industrial standards within the PA domain are mainly based on the PHY layer of the IEEE 802.15.4 standard. The PHY layer of the IEEE 802.15.4 standard is designed to support short-range communications that offers a low data rate, low-cost and low-power consumption solution for industrial applications [27]. A rough comparison between the standards is shown in Fig. 2.1.

2.2.1 WirelessHART

WirelessHART was ofﬁcially released by the HCF in 2007, aiming to be compatible with existing HART devices by adding wireless communication capability to the 2.2 Overview of the Industrial Standards for PA Applications 9

Figure 2.1: Protocol stack comparison of industrial standards for PA applications.

HART protocol in PA and control scenarios. WirelessHART operates at the 2.4 GHz band and supports up to 15 channels. It utilizes the Time Synchronized Mesh Pro- tocol (TSMP) [28] for MAC and network layer functions. TSMP uses Time Division Multiple Access (TDMA) for channel access and allows for channel hopping and channel blacklisting at the network layer. Channel hopping is a technique in which data transfer happens at different frequencies at different periods of time. Channel blacklisting is a process of rejecting (blacklisting) channels that exhibit large interference levels.

2.2.2 ISA100.11a

The ISA group established the ISA100.11a standard in 2009, mainly aiming for a robust and secure communications for PA applications. Unlike WirelessHART, ISA100.11a applies different mechanisms for channel hopping, such as slotted, slow and hybrid channel hopping in order to avoid collision with surrounding IEEE 802.11 networks. In addition, ISA100.11a includes backbone routers for bridging subnets while WirelessHART uses access points. The compatibility of ISA100.11a with IPv6 in the network layer gives the users the opportunity to connect to the internet, thus providing more options. ISA100.11a supports star and mesh network topologies and offers an interface for the integration with WirelessHART [29].

2.2.3 WIA-PA

WIA-PA standard was established in 2007 with the aim of providing energy efﬁcient, highly reliable and intelligent multi-hop IWSNs that are more reactive to dynamic change in the network. It has 16 channels in the 2.4 GHz band with three different modes of frequency hopping, slotted channel hopping, adaptive channel hopping and adaptive frequency switch. In contrast to WirelessHART and 10 Background and Related Work

ISA100.11a, WIA-PA adopts the IEEE 802.15.4 MAC layer without modification in order to easily co-exist with extensive existing IEEE 802.15.4-based systems. An- other difference is that WIA-PA network supports a hierarchical network topology, which is basically a star-plus-mesh topology. The first level of the network has a mesh topology, which consists of routers and gateways, while the second level is a star topology, composed of routers and field devices. WIA-PA supports interoper- ability with legacy protocols, such as wired HART and PROFIBUS, and also offers support for WirelessHART.

2.2.4 Zigbee

The Zigbee wireless standard was developed for a wide range of applications, including industrial automation, however not specifically designed for industrial applications. It is mainly based on the IEEE 802.15.4 specifications and is mainly suitable for applications where low-power consumption is given more importance than providing a real-time performance. Network topologies supported by ZigBee are star, tree, and mesh topologies, with the latter resulting in a highly dynamic and distributed system. The number of channels that ZigBee can use is 27, since it utilizes all three bands defined by the PHY layer in IEEE 802.15.4.

2.2.5 IEEE 802.15.4e

Due to the shortcoming of the IEEE 802.15.4 MAC for industrial applications, the IEEE 802.15.4e working group was created in 2008 to enhance the IEEE 802.15.4 MAC protocol, in order to provide improved support for PA applications [10]. The IEEE 802.15.4e standard offers two operation modes for PA applications, which are Time Slotted Channel Hopping (TSCH) and Deterministic and Synchronous Multi- channel Extension (DSME) [30]. The core technology of the TSCH is mainly inherited from WirelessHART and ISA100.11a standards (combined TDMA and channel hopping feature) [31], while the key technology of the DSME is also adopted by WIA-PA standard in advance. The operation of each mode is mutually exclusive and the operator is recommended to select one of operation modes to form a network.

2.3 MAC Protocols in IWSNs

The MAC protocol is fundamentally responsible for controlling the access to the shared medium and the underlying communication scheduling among sensor nodes, which in turn controls the data transmission capability. Hence, a necessary ﬁrst step to support time- and mission-critical applications is to ﬁnd a MAC protocol capable of satisfying deterministic performance bounds regarding delay and reliability. 2.3 MAC Protocols in IWSNs 11

Figure 2.2: Wireless MAC protocols taxonomy.

2.3.1 General Taxonomy of Wireless MAC Protocols

MAC protocols for wireless technology can be roughly classiﬁed into the categories shown in Fig. 2.2 [32]:

• Fixed-Assignment Protocols: the available resources are divided appropriately among the nodes for a long term. TDMA, Frequency Division Multiple Ac- cess (FDMA) and Code Division Multiple Access (CDMA) are examples of this protocol type. In TDMA, each node is pre-scheduled for channel access by assigning an exclusive time slot for its transmission. The pre-deﬁned schedule is repeated periodically based on the data refresh rate. TDMA has been widely utilized in IWSNs since it is able to provide a guaranteed channel access and offers low complexity requirements in transceivers and micro-controllers. In FDMA, the available band is divided into sub-channels and each node is assigned a dedicated sub-channel, while in CDMA the signals are sent using spread spectrum technology, and a special encoding scheme is required to transmit signals in the same channel at the same time.

• Random Access Protocols: the medium is available for all nodes at the same time and each node should compete to gain channel access. Slotted-ALOHA and Carrier Sense Multiple Access with Collision Avoidance (CSMA/CA) are common representatives of this protocol type. In slotted-ALOHA, time is equally sliced into discrete transaction slots and a node can transmit its data in the beginning of the next slot whenever it has a packet to send. CSMA/CA improves slotted ALOHA by introducing a channel sensing scheme to detect the channel availability using the CCA approach. In CSMA/CA [16], a node 12 Background and Related Work

that has a packet to transmit starts by sensing the channel, and if it is found to be idle, then the packet is transmitted. If the channel is busy, the node defers its transmission for a random back-off time, which is uniformly selected from the range [0, 320(2BE − 1)µs], where BE is the back-off exponent. The CSMA/CA algorithm is implemented using units of time called back-off periods [16].

• Demand Assignment Protocols: resources are temporarily allocated to nodes. Once the transmission is completed, the resources are returned. The protocols can be used either in centralized or distributed modes. Token-passing protocols are a well-known example of the distributed demand assignment protocols.

• Hybrid Protocols: combining different types of the aforementioned protocols to exploit their advantages and avoid the limitations. One example is the Z-MAC protocol [33], which combines TDMA and CSMA/CA in order to achieve high throughput under low and high contention levels.

2.3.2 MAC Protocols in IWSNs Standards

Generally speaking, the MAC function in the available PA standards adopts TDMA, CSMA/CA, slotted-ALOHA, or a combination of them to schedule data transmissions through the shared medium.

• WirelessHART/ISA100.11a MAC: The superframe structure for both WirelessHART and ISA100.11a is depicted in Fig. 2.3(c). The superframe contains two types of slots, dedicated time slots and shared time slots. The dedicated time slots are based on TDMA, where the network manager assigns a scheduled time slot to transmit a message from one node to another. Transmis- sions in the shared time slots are scheduled based on slotted-ALOHA. Shared slots are used for join, advertise and retransmission trafﬁc. The superframe in WirelessHART is a distinctive collection of exactly 10 ms time slots, while the superframe in ISA100.11a can be a hybrid of short and long time slots in the interval 10 − 12 ms.

• WIA-PA MAC: WIA-PA employs a compatible superframe structure with Zigbee, where the CAP and the CFP parts have redefined purposes as shown in Fig. 2.3(b). The CAP period of WIA-PA is used for device joining, intra-cluster management and retransmissions, while the CFP is used for communication between nodes and the cluster head. The WIA-PA time slot duration in CFP is configurable and compatible with Zigbee GTS. The inactive period defined in Zigbee superframe is modified in the WIA-PA for intra-cluster, and inter- cluster communications, as well as a sleeping period.

• Zigbee MAC: The Zigbee uses a hybrid approach for CSMA/CA and TDMA to manage data transmissions. Fig. 2.3(a) shows the superframe structure in Zigbee. Each superframe has active and inactive periods. The active period is composed off three parts, beacon, Contention Access Period (CAP) and 2.3 MAC Protocols in IWSNs 13

Figure 2.3: The superframe structure of Zigbee, WIA-PA, WirelessHART and ISA100.11a [9].

Contention Free Period (CFP). The coordinator maintains the synchronization through the transmission of periodic beacons. Following the beacon, the nodes contend for channel access during the CAP using CSMA/CA approach. The CFP follows the CAP and extends to the end of the active period. In the CFP, the coordinator centrally assigns Guaranteed Time Slots (GTSs) to allow a deterministic data transmission. Up to 7 GTSs can be assigned to the devices in each superframe.

• IEEE 802.15.4e MAC: Two modes of operation are adopted by the MAC function in IEEE 802.15.4e, TSCH and DSME. The core technologies of the TSCH originates from WirelessHART [34], while the key technology of the DSME is also adopted by WIA-PA in advance. TSCH combines the TDMA approach along with multi-channel and channel hopping approaches. It relies on cyclic slotframes that are comprised of a ﬁxed number of time slots. The slotframe can be represented by a scheduling matrix as shown in Fig. 2.4. Each cell is deﬁned by a pair of time slots and channel offset denoting respectively its in- 14 Background and Related Work

Figure 2.4: TSCH superframe.

Figure 2.5: DSME multi-superframe structure.

stant of transmission in the slotframe and the frequency for the transmission. As shown in Fig. 2.5, DSME uses a versatile multi-superframe structure that extends the number of GTSs and increases the number of frequency channels used.

Despite the potential of the above standards in a wide range of industrial applications, they cannot support the strict real-time requirements of time- and mission- critical applications in a satisfactory manner. The random access methods CSMA/CA and slotted-ALOHA utilized in the above standards offer some advantages, such as easy implementation; full bandwidth being available to any node; and suitability for use in dynamic networks without pre-scheduling. However, these protocols are unable to guarantee deterministic performance for time- and mission- critical applications because of the unpredictable time delay being generated by the random backoff [35]. In addition, these random access methods cannot provide a prioritization mechanism based on timing criticality of the trafﬁc being delivered. Although the scheduled-based approaches, such as GTS and TDMA, adopted by these standards can provide more deterministic performance compared to the random access protocols, they are not good choices for time- and mission-critical applications in IWSNs. TDMA mainly follows a ﬁxed pre-scheduled slotted channel access scheme, hence it is unable to guarantee immediate channel access for the non-predictable time-critical data (e.g., emergency data) due to its non-deterministic 2.3 MAC Protocols in IWSNs 15

occurrence. Accordingly, the transmission of such data should wait for its dedicated time slot, which is not acceptable for industrial applications with strict deadlines. The problem becomes even worse for dense networks where the TDMA frame consists of hundreds of scheduled time slots, which also leads to scalability issues. In addition, if a time slot is reserved for emergency data in each frame, as this type of data does not occur periodically. The GTS mechanism can provide a number of deterministic transmission opportunities for nodes with critical data, however, this mechanism suffers from weak points. First, the number of GTS is limited to seven (GTS starvation) [36], i.e., the network cannot support a guaranteed access for many devices. Second, before a node uses a GTS, it has to send a GTS request in the CAP and wait for the allocation confirmation in the next beacon interval. Third, the GTS allocation is performed based on the First-Come-First-Served (FCFS) approach, which can cause the critical traffic to be blocked when all the GTSs are allocated to the non-critical traffic.

2.3.3 Related Work

Significant research work has been done in the literature to design MAC protocols where the main design goal is to improve the energy efficiency of the WSN to maintain a long lifetime of the network and reduce maintenance cycles [37]. These protocols are very energy efficient, however, they provide only simple best-effort delay performance for data delivery. Among them, S-MAC protocol [38] is one of the prominent pioneering studies on MAC protocols in the context of energy efficiency, where the periodic duty-cycle concept is utilized to reduce idle listening. A node co- ordinates its fixed sleep/active period with neighbours using SYNC packets. Within an active period, nodes follow the Request-To-Send (RTS) and Clear-To-Send (CTS) handshake mechanism to avoid collision. However, because of the scheduling of the fixed duty cycle, this will cause an unacceptable delay, which makes this MAC protocol unsuitable for time- and mission-critical applications. The reliability and timing constraints required in such applications can be met by introducing routing protocols [39] or MAC protocols that are capable of satisfying the stringent real- time performance. Focusing on MAC protocols, several MAC protocols have been proposed in the literature in order to overcome the aforementioned limitations and provide real-time communication for time- and mission-critical applications [24]. Adaptive scheduling and allocation algorithms are introduced in [40–43] to provide low latency and efficient utilization of time-critical data. However, these algorithms mostly adopt CSMA/CA for channel access, which cannot provide the deterministic delay performance required by time- and mission-critical applications. Furthermore, no priority differentiation mechanism is considered to prioritize critical data over the non-critical one. A priority-aware multi-channel framework has been proposed in [44] to improve the Low-Latency Deterministic Network (LLDN) in IEEE 802.15.4e by introducing a message priority mechanism with multi-channel transmission to avoid deadline miss. The work only considers periodic traffic, which is not the case for critical events generated at unpredictable instants. The HyMAC protocol [45] is a hybrid TDMA/FDMA protocol dividing time into 16 Background and Related Work

fixed-length frames; each frame is further divided into either scheduled time slots for collision-free transmission, or contention time slots for transmitting control messages. Each node utilizes the contention time slots to send control messages to the base station, including a list of its neighbours. The base station utilizes this list to construct a tree topology and minimum delay schedule for node transmissions by assigning the appropriate time slot and frequency channel. Although HyMAC guarantees a certain end-to-end delay, there is no priority mechanism provided to enable timely delivery of the critical data. The authors in [46] introduced a hybrid TDMA/CSMA protocol, named ER- MAC, for emergency response in WSNs. Each nodes utilizes two queues for high and low priority traffic. In normal node, data transmissions are scheduled based on TDMA, where each node is assigned a time slot, while in emergency mode, the nodes that are contributing in the emergency event utilize CSMA/CA to contend for access to the TDMA slots after exchange request messages with the slot owner. However, as mentioned earlier, CSMA/CA is not a suitable approach for transmitting emergency messages as it cannot provide the required predictable delay performance. Zhang et al. [47] proposed a distributed dynamic packet scheduling algorithm, referred to D2-PaS, to handle critical events caused by external disturbances in real- time WSNs. In order to guarantee the delivery of the critical events by their deadlines, the proposed scheduling algorithm forces the medium to be idle for a number of consecutive time slots, which introduces a significant bandwidth wastage if there is a large number of nodes. The authors in [48] presented the GinMAC protocol to provide timely delivery for time-critical applications by using exclusive time slots for transmission. Since GinMAC mainly utilizes TDMA approach, it is unable to guarantee immediate channel access for emergency data and there is no mechanism provided to prioritize channel access for critical-time data. Another drawback of GinMAC is its limited scalability as it is designed to support a network with a max- imum of 25 nodes. The concept of slot stealing is adopted by some works in order to improve the transmission efficiency of acyclic traffic in WSNs [49–51]. Li et al. [49] proposed a real-time communication protocol to incorporate emergency alarms in WSNs. In the proposed protocol, emergency nodes are pre-assigned fixed time slots for emergency traffic transmission and these slots can be stolen by regular nodes when there is no emergency. However, no analysis to derive the delay bound for the emergency transmission is provided. The authors in [50] introduce a slot stealing strategy to guarantee real-time performance and reliability for mixed-criticality systems, where high-critical flows are enabled to steal time slots from low-critical ones. The work has been extended in [51] by introducing a series of algorithms to determine the optimal path for the transmission of the unpredictable emergency events. The authors in [52] presented a PriorityMAC protocol that considers four priority levels for data transmission. The proposed protocol provides timely access for the highest priority data by hijacking the dedicated transmission bandwidth of the lower priority data. The channel access scheme utilized by the PriotiyMAC cannot provide a deterministic channel access for time-critical data due to random delays caused by collisions. 2.3 MAC Protocols in IWSNs 17

Furthermore, none of the above solutions define an efficient scheduling mechanism for channel access in case an emergency event triggers multiple sensor nodes to simultaneously send time-critical data to the controller with different deadline bounds. A novel MAC protocol, named WirArb, is proposed in [53] to enable deterministic channel access for critical data in a fashion similar to the aforementioned CAN protocol. In the proposed protocol, each user is assigned a pre-defined arbitration frequency to deterministically prioritize channel access, which ensures that the user with the lowest frequency gains immediate channel access. The proposed protocol is still in its early stages and needs a complete redefinition regarding the PHY layer modelling of the gateway and sensor nodes. Motivated by the gaps identified in the above literature work, this thesis aims to develop efficient MAC protocols to enable time- and mission-critical applications in IWSNs. 18 Chapter 3

Priority-Based Real-Time Communication in IWSNs

3.1 Overview

As discussed in the previous chapters, providing real-time communication for time- and mission-critical applications is a fundamental research problem in the industrial domain. This problem is approached in this work by proposing MAC protocols that enable a timely delivery of unpredictable critical traffic over the non-critical traffic through priority-based channel access schemes. The work started by modelling the effect of priority-based transmissions to evaluate the average delay performance in low and heavy traffic conditions. Then, two MAC protocols are to enable timely and prioritized channel access of the critical traffic. The following sections describe approach proposed to achieve the research objective of this thesis.

3.2 Modelling Priority-Oriented Packet Transmissions

First, the effect of introducing priority transmission on the average delay performance in low as well as heavy trafﬁc conditions is studied. The priority-based transmission is applied in the node level concept, i.e., each node utilizes two queues in its output buffer, low priority queue and high priority queue, where each packet is placed in its corresponding queue based on its type. We developed a queuing- based two dimensional DTMC model. The developed 2D DTMC models the nature of a priority-based transmissions that is employed over a synchronous duty cycling MAC protocol, S-MAC [38]. The two dimensions in the model correspond to number of packets in a low priority queue and high priority queue. Furthermore, to model the number of active nodes in the network, a 1D DTMC is developed. The two models are fused jointly to investigate the average delay in packet transmissions. Un- like existing models for synchronous DC MAC protocols in [54–56], the developed

19 20 Priority-Based Real-Time Communication in IWSNs

DTMC models embody the feature of priority aware packet transmissions. Considering a network of N nodes, to compete for access to the medium, the contending nodes generate random back-off time from the set {0,W − 1} during the active period. The successful node is the one that selects the smallest back-off time of the contending m, 0 ≤ m ≤ N − 1, active nodes. The successful probability can be given as W −1 X m m Ps,m = (1/W )(W − 1 − i) /W , (3.1) i=0

A state in the 2D DTMC is represented by (i, k), where i is the number of packets available in the low priority queue and k is the number of packets available in the high priority queue of a particular node. A node transmits low priority packets only if there are no high priority packets ready for transmission . Otherwise, high priority packets would be transmitted. Considering that the arrival of high priority and low priority packets to both queues follow a Poisson process, with rates λhp and λlp, respectively, then a transition in the 2D DTMC occurs based on the successful transmission probability, λhp and λlp. A packet, either low priority or high priority, will stay in the buffer until it is successfully deliverd. Hence, the delay is deﬁned as the duration elapsed until the packet is delivered after its entry into the buffer. According to Little’s law, the average delay of a corresponding packet D, is given by

Nav D = , (3.2) γa where Nav is the average number of packets present in a corresponding queue and γa is the rate at which the packets joins the queue. Nav can be formulated using the developed 2D DTMC model as follows

Q X Nav = iπi, (3.3) i=0 where πi is stationary distribution of the 2D DTMC. More details on the developed DTMC models can be found in Paper I. The priority-oriented packet transmission scheme along with the S-MAC protocol is simulated in a discrete-event simulator to verify the analytical models. Further- more, the simulation results are averaged over 5 × 106 cycles. Apart from the good matching of the numerical and discrete-event simulation results shown in Fig. 3.1a , it can be observed that the non-emergency data experiences insignificant average delay in the absence of emergency traffic. When coexisting with high priority traffic, non-critical packets are delivered very late. Moreover, Fig. 3.1b shows the effect of the number of nodes N on the average delay. It can be seen that when the high priority traffic is not considered, the delay in the non-critical traffic with growing N increases minimally. However, in the presence of high priority traffic together with a higher access contention, the low priority packets rarely receive transmission opportunities and thereby suffer from an extremely long delay. 3.3 Improving the Transmission Efficiency of Acyclic Traffic in IWSNs 21

300 500 LP, λ =0.5 packet/s, analysis LP, λ =0.5 packet/s, analysis hp hp 250 LP, λ =0.0 packet/s, analysis LP, λ =0.0 packet/s, analysis hp 400 hp LP, λ =0.5 packet/s, simulation LP, λ =0.5 packet/s, simulation hp hp 200 LP, λ =0.0 packet/s, simulation LP, λ =0.0 packet/s, simulation hp hp 300 150

D (cycles) 10 200

100 D (cycles) 2 5 1.5 50 0 100 1 0.8 0.9 1 18 19 20

0 0 0 0.2 0.4 0.6 0.8 1 10 12 14 16 18 20 λ (packets/s) lp N

(a) Delay at N = 15 as λlp varies. (b) Delay at λlp = 0.5 as N varies.

Figure 3.1: Performance comparisons.

Figure 3.2: Slot timing structure of the proposed PR-CCA protocol.

3.3 Improving the Transmission Efﬁciency of Acyclic Trafﬁc in IWSNs

Next, the priority-based transmission is extended to the network level concept with the goal of enhancing the transmission efficiency of the acyclic traffic in terms of throughput and delay. The unpredictability regarding the arrival of this traffic (or its retransmission instants) imposes a key challenge in scheduling its transmission since it is difficult to construct a suitable schedule or reserve a dedicated bandwidth to guarantee its timing constraints. In this context, PR-CCA protocol is introduced, a prioritized random CCA-based MAC protocol to enhance the transmission efficiency of event-based traffic in industrial applications. In PR-CCA, acyclic critical traffic is given a higher priority and is allowed to opportunistically access the time slots assigned for the lower priority traffic using an efficient CCA mechanism. A number of CCA subslots are inserted in each time slot as depicted in Fig. 3.2 where each node performs the CCA for a randomly selected number of subslots. This way, the collision probability is significantly decreased, which in turn improves the per- 22 Priority-Based Real-Time Communication in IWSNs

(a) Average delay comparison. (b) Reliability comparison.

Figure 3.3: Performance comparisons. formance in terms of delay and throughput. Moreover, a DTMC model is developed to evaluate the performance of the proposed protocol in terms of throughput and average delay. More details on PR-CCA protocol and the developed DTMC model can be found in Paper II. Numerical results have been given by evaluating the mathematical equations developed in Paper II to introduce performance comparisons between PR-CCA and WirelessHART. Figure 3.3 shows that introducing CCA subslots to decrease the contention in acyclic trafﬁc noticeably improves the overall performance regarding average delay (Fig. 3.3a) and reliability (Fig. 3.3b). In addition, the ﬁgure proves the performance improvements compared to the WirlessHART standard.

3.4 Deterministic Real-Time Communication of Multi- ple Critical Flows

The emerging IWSNs are expected to satisfy the deterministic performance that is met by wired solutions, in particular, this applies to time- and mission-critical applications. In this context, the MAC protocol should provide a delay guarantee for critical data delivery for which function cannot be ensured by the contention-based MAC protocols. Existing industrial standards can only provide the deterministic performance for the periodic traffic by adopting conventional TDMA scheduling. However, due to the unpredictable nature of the acyclic critical traffic, it is difficult to schedule guaranteed transmission opportunity for such traffic. Also, it is more challenging for emergency situations where multiple sensor nodes are simultaneously triggered to send critical data to the controller with different deadline bounds, for instance, in a fire emergency where a combination of sensors, e.g., smoke, temperature and CO can collaborate to detect the presence of fire when the sensor reading is above a specified threshold. In this context, SS-MAC is proposed, a slot stealing MAC protocol that provides 3.4 Deterministic Real-Time Communication of Multiple Critical Flows 23

Figure 3.4: SS-MAC channel access scenario. deterministic and predictable real-time communication for time- and mission-critical applications in IWSNs. The triggered nodes are deterministically scheduled by the controller node to transmit their critical data based on a dynamic deadline-aware schedule; that is the node with the most urgent data obtains the highest priority and gains immediate channel access. A Reservation Request Phase (RRP) is inserted in each time slot to enable each critical node to reserve its transmission opportunity and declare its priority based on its corresponding deadline. At the end of the RRP, the controller broadcasts the channel access schedule, which includes a Channel Access Order (CAO) for each triggered node. The node with the highest priority, i.e. closest deadline, then transmits its critical data to the controller. An illustrative scenario for the proposed SS-MAC is shown in Fig. 3.4. In the depicted scenario, the nodes n1, n2 and n3 are simultaneously triggered to send critical data to the controller within deadlines d1, d2 and d3, respectively. For simplicity, the deadlines are assumed to be in ascending order, i.e., d1 < d2 < d3. Each node transmits its reservation packet within its corresponding subslot during the RRP. The controller constructs the schedule based on the received deadlines, i.e., earliest due date scheduling, and broadcasts the CAO values. Since n1 has the earliest deadline, it gains the highest priority by receiving CAO = 0 and transmits its critical data to the controller accordingly. The lower priority nodes n2 and n3 receive CAO = 1 and CAO = 2, respectively and de- fer their transmissions to the following transmission cycles. The proposed protocol has been evaluated mathematically to provide the worst-case delay bound for the time-critical trafﬁc, which is given as

Dworst = KTss + TRxCAO + Tdata + (K − 1)(Ts + TEIS), (3.4) where Dworst is the worst-case delay, K is the number of critical nodes in the network, Tss is the duration of one subslot in the RRP, TRxCAO is the time to receive the CAO, Tdata is the duration of the transmission of the critical packet, Ts is the time slot duration and TEIS is the duration of an emergency indication subslot. The performance of the proposed SS-MAC is evaluated numerically using MAT- LAB in terms of delay and channel utilization and is compared with a TDMA-based system, such as WirelessHART. The results are obtained using the values and param- 24 Priority-Based Real-Time Communication in IWSNs

Table 3.1: System Parameters [57].

Parameter Value Transmission rate 250 kbps Tss 352 µs TRxCAO 640 µs Ts 10 ms TEIS 320 µs PHY header 192 µs MAC header 224 µs eters listed in Table 3.1. Transmission rate, PHY header, and MAC header are mainly adopted from the IEEE 802.15.4 standard. The value of TEIS is set according to the minimum time unit duration in the IEEE 802.15.4 standard, which equals to 320µs. Tss is set according to the considered size of the reservation request frame. The reservation request frame contains 2 bytes (frame control), 2 bytes (source address) and 1 byte (payload). With 6 bytes (PHY header), Tss is calculated as

11 bytes × 8 T = = 352 µs. (3.5) ss 250 kbps

TRxCAO in Table 3.1 is equivalent to the duration of the transmitted broadcast signal from the controller that contains a list of the CAO values. This can be viewed as transmitting the beacon frame in the IEEE 802.15.4 standard, which includes a field containing the information about the reserved GTS in the superframe. We assume that the frame of the broadcast signal only contains a frame control field, and a payload field, which includes the list of the CAO values. Setting the size of the payload field to 12 bytes, which is reasonably sufficient compared to the GTS field in the beacon frame of the IEEE 802.15.4 standard, and with 2 bytes (frame control) and 6 bytes (PHY header), then 20 bytes × 8 T = = 640 µs. (3.6) RxCAO 250 kbps

The value of Ts is set in order to maintain compatibility with the WirelessHART time slot duration, which is speciﬁed in the standard to equal 10 ms. Figure 3.5 shows the worst-case delay comparisons of the critical trafﬁc between the proposed SS-MAC and a TDMA-based WSN such as WireleesHART under the different values of the number of nodes. The worst-case delay for the proposed SS-MAC is evaluated under different values of the critical nodes K in the network, which are represented as a fraction of the total number of nodes N, e.g, K = N/2, N/3 and N/4 and the payload size of the critical packet is 100 bytes. For the TDMA system with a total of N nodes (N time slots), if the delay calculation is considered from the boundary of the time slot, then the worst-case delay for a certain critical node can be given as

DTDMA worst = (N − 1)Ts + Tdata. (3.7) 3.4 Deterministic Real-Time Communication of Multiple Critical Flows 25

Figure 3.5: Worst-case delay comparison versus different number of nodes.

(a) Number of nodes N = 20 and superframe (b) Number of nodes N = 20 and number of length x = 10. critical nodes K = 15.

Figure 3.6: Utilization comparisons.

As shown in Fig. 3.5, the proposed SS-MAC provides enhanced worst-case delay performance compared to the TDMA-based channel access mechanism for different values of K, for example, the SS-MAC can decrease the worst-case delay of the TDMA by about 48% for N = 20 and K = 10. This is because the efficient on- demand channel allocation strategy utilized by the SS-MAC, which is based on a time-criticality priority differentiation that enables the highest priority node to gain immediate channel access for urgent data transmission. However, TDMA is mainly based on a fixed time slot allocation and the transmission of the critical data has to wait for its dedicated time slot, resulting in increased transmission delay which is unacceptable in many time-critical applications with strict deadlines. Figure 3.6 shows the utilization comparisons between the TDMA scheme and the proposed SS-MAC versus the number of critical nodes (Fig. 3.6a) and the superframe length (Fig. 3.6b), respectively. The channel utilization is defined as the percentage 26 Priority-Based Real-Time Communication in IWSNs

of the total number of transmissions per time slot in a superframe, which can be expressed as follows PN E[ Si] U = i=1 , (3.8) E[ST ] where Si represents the total number of transmissions by ni in the superframe and ST is the total number of time slots in the superframe. These ﬁgures clearly show that SS-MAC can achieve 100% channel utilization regardless of the number of critical nodes or the length of the superframe. This is because of the dual-reservation mechanism utilized by the proposed SS-MAC, in which the time slots are always fully exploited by either critical or non-critical data transmissions. However, the TDMA scheme provides poor channel utilization due to the ﬁxed reservation of periodic time slots for the critical transmissions, which occasionally occur and the performance is further degraded with the increase of the number of critical nodes and the superframe size as more time slots are wasted. For example, while the proposed SS-MAC maintains a steady-full channel utilization as K or x increases from 1 to 15, the channel utilization in WirelessHART is degraded by 60%. Chapter 4

Summary of Publications

This chapter summarizes the papers associated with this thesis and presents some limitations that are not covered in the papers.

Paper I Priority-Oriented Packet Transmissions in In- ternet of Things: Modeling and Delay Analysis

Motivation and Contributions In WSNs designed for smart cities or industrial applications, packets from surveil- lance or monitoring services are sent regularly. In addition, critical detection or alarm packets are rarely generated in emergency situations. Control packets indi- cating emergencies need to be transmitted immediately to promote taking action before events lead to a disaster. Therefore, a higher priority should be given to critical packets that are sensitive to delay while delay-tolerant messages can be transmitted when there is no time-critical trafﬁc. To this end, the effect of trafﬁc prioritization needs to be analysed for WSNs. This issue was highlighted by RQ1 and addressed by the work presented in Paper I. In this paper we evaluate the performance of a priority-oriented packet transmissions scheme with two priority levels, low priority and high priority, using two associated DTMC models. The developed 2D DTMC models the nature of a priority- based transmission scheme that is employed over a synchronous duty cycling MAC protocol, S-MAC. The two dimensions in the model correspond to the number of packets in a lower priority queue and a higher priority queue. Furthermore, to model

27 28 Summary of Publications

the number of active nodes in the network, a 1D DTMC is developed. These two DTMC models are fused to investigate the effect of introducing priorities in packet transmissions in terms of average delay performance.

Novelty The idea of packet prioritization has been investigated in many works where a particular packet is assigned a priority based on its delivery deadline. Unlike existing models for synchronous duty cycling MAC protocols, in this work, the effect of introducing priorities in packet transmission is analysed using two DTMC models that mathematically embody the priority feature to evaluate the network performance in terms of average delay. In addition, our models analyse the network characteristics in both low as well as heavy trafﬁc conditions.

Limitations Although the proposed mechanism introduces lower latency for the higher priority packets compared to the lower priority ones, it cannot guarantee that a higher priority packet is always transmitted ﬁrst in the network scale sense, i.e., a node that has a lower priority packet may win the contention while another node with a higher priority packet fails and waits for the next time contention phase. Additionally, the analytical model needs to be extended to formulate the worst-case delay of higher priority packets.

Author Contribution Hossam Farag contributed in the problem description, formulating the protocol mechanism and paper writing.

Paper II PR-CCA MAC: A Prioritized Random CCA MAC Protocol for Mission-Critical IoT Applications

Motivation and Contributions Typical control applications in the industrial automation domain includes acyclic critical traffic, which is generated at unpredictable time instants as a result of a certain observation of an established tendency, such as closed-loop supervisory traffic. This type of traffic is delay sensitive and if it arrives too late, it is of limited use and may cause degradation in the control system, economical loss, or even worse, Summary of Publications 29

endanger human life. Due to its non-deterministic occurrence, scheduling the transmissions of such traffic in a delay-efficient way and its prioritization over regular traffic are key challenges in the design of IWSNs. Typical industrial standards such as WirelessHART and ISA100.11a cannot handle the transmissions of such traffic in an efficient way. The aforementioned issue is raised in RQ2, which is investigated in Paper II. In this paper, we propose a prioritized random CCA-based MAC protocol to enhance the transmission efficiency of acyclic traffic in industrial control applications. In PR-CCA, acyclic critical traffic is given a higher priority and is allowed to opportunistically access the time slots assigned for lower priority traffic using a novel CCA mechanism. This way, the collision probability is significantly decreased which in turn improves the performance in terms of latency and throughput. We develop a DTMC model to evaluate the performance of the proposed protocol. The results show performance improvements of 80% and 190% of the expected delay and throughput, respectively, compared with the WirelessHART standard along with better transmission reliability. In addition, the proposed PR-CCA protocol is compliant with the slot timing structure of WirelessHART.

Novelty PR-CCA proposes a novel contention-based mechanism that significantly en- hances the transmission/retransmission efficiency of acyclic control traffic in terms of both throughput and delay. Unlike the conventional fixed CCA mechanism adopted in most existing works, PR-CCA introduces a random selector for the CCA period, which mainly mitigates the collision probability of contending nodes, if not eliminated completely, when more than one node choose the same CCA slot to perform channel sensing. The introduced CCA period can be inserted directly in TsTx- Offset, which makes PR-CCA easily compliant with the MAC of WirelessHART and ISA100.11a.

Limitations The proposed protocol lacks simulation and/or real-world experiment evaluations to further validate the theoretical analysis. In addition, the performance of the proposed work needs to be investigated under multi-hop scenarios and hidden-node problem.

Author Contribution Hossam Farag is the main author of this article and responsible for problem description, protocol idea and mathematical evaluations. 30 Summary of Publications

Paper III A Delay-Bounded MAC Protocol for Mission- and Time-Critical Applications in Industrial Wire- less Sensor Networks

Motivation and Contributions IWSNs designed for time- and mission-critical applications require timely and deterministic data delivery within stringent deadline bounds. This means that the transmission of critical traffic, such as safety-related data, should be characterized by a deterministic delay bound. In addition, another vital challenge is considered to schedule the transmissions in emergency situations where multiple sensor nodes are simultaneously triggered to send critical data to the controller within different deadline bounds. To this end, RQ3 is addressed in Paper III. In this work, we propose SS-MAC, an efficient MAC protocol to guarantee deterministic real-time communication for time-critical data in IWSNs. SS-MAC pri- oritizes the transmission of critical data by enabling a slot stealing mechanism, in which the aperiodic critical traffic opportunistically steals time slots assigned to the periodic non-critical traffic. In addition, the triggered nodes are deterministically scheduled by the controller node to transmit critical data based on a dynamic deadline-aware schedule; that is, the node with the most urgent data (closest deadline) obtains the highest priority and gains immediate channel access. The proposed protocol is evaluated mathematically in terms of the transmission delay to give the upper bound for the worst-case delay. Furthermore, the performance of SS-MAC is compared with TDMA-based WSNs such as WireleesHART and the results show that the proposed protocol can provide a superior performance for time-critical applications in terms of the worst-case delay and channel utilization.

Novelty The slot stealing mechanism has been addressed in many works in the literature, however, SS-MAC introduces a number of novel features. First, SS-MAC enables guaranteed channel access for critical trafﬁc transmission by adopting a deadline- aware scheduling algorithm that ensures immediate channel access for the most urgent node. Second, while most of the existing works consider scheduling only a single critical ﬂow at a time, SS-MAC deterministically schedules the transmission of multiple critical nodes that are triggered simultaneously during a particular emergency situation. Third, SS-MAC guarantees the required deterministic performance in time- and mission-critical applications by providing an upper bound for the worst-case delay of critical data, while previous works in the context of slot stealing mostly evaluate the performance in terms of the average delay and overlook the worst-case delay guarantee. Summary of Publications 31

Limitations The number of emergency nodes that can be handled is the main limitation of the proposed solution. Using SS-MAC in large scale networks is not efﬁcient as it may introduce high latency for critical nodes. In addition, performance assessments under multi-hop scenarios and synchronization issues are not presented in this article. Also, the reliability of the proposed protocol under harsh channel conditions is not considered, which has a direct effect on the delay performance.

Author Contribution Hossam Farag is the main author of this article and responsible for problem description, protocol idea and mathematical evaluations. 32 Chapter 5

Conclusions and Future Work

Existing industrial wireless standards within the PA domain are unable to support time- and mission-critical applications due to their limited functionality to handle the delivery of the unpredictable critical trafﬁc in a timely fashion. This thesis aims to address vital challenges in this context regarding providing real-time communication and service differentiation for medium access. This work contributes with MAC-based solutions that can be integrated with existing industrial standards to enable real-time communications in time- and mission-critical applications. The following section reviews the link between the identiﬁed research questions and the contribution presented in this work.

5.1 Concluding Remarks

First, the issue of introducing service differentiation and how it affects the average delay performance of the network is addressed by RQ1. In this context, in Paper I, the performance of priority-oriented packet transmissions in a duty-cycled MAC is analysed using two DTMC models. The developed DTMC models embody the feature of priority aware packet transmissions to evaluate the average delay performance in low and heavy traffic conditions. The analytical results are verified through discrete-event simulations to prove the accuracy of the models as well as evaluate the behaviour of priority-based packet transmissions. In order to further improve the transmission efficiency of acyclic critical data in control applications, a PR-CCA protocol is proposed and formalized in Paper II with the goal to address the challenges raised in RQ2. PR-CCA introduces a priority- based random CCA mechanism to handle the simultaneous transmissions of acyclic critical data and reduce the collision probability between the contending nodes. The performance of the proposed approach is evaluated using a DTMC model and the obtained results show that the proposed protocol can achieve significant improvements regarding the expected delay and throughput compared with the

33 34 Conclusions and Future Work

WirelessHART standard along with better transmission reliability. For critical applications that are characterized by strict deadlines, bounded and guaranteed delay performance is a must. It also becomes more challenging when multiple nodes are simultaneously triggered to transmit critical data to the sink. The aforementioned challenges are addressed in RQ3 and the corresponding solution is proposed in Paper III through the SS-MAC protocol. SS-MAC presents a slot- stealing MAC protocol to handle critical data communications in emergency and safety-related situations. The triggered critical nodes are deterministically scheduled by the controller node to transmit their critical data based on a dynamic deadline- aware schedule; that is, the node with the most urgent data obtains the highest priority and gains immediate channel access. SS-MAC provides a deterministic delay guarantee performance and the obtained numerical results demonstrate that SS- MAC attains better performance in terms of the worst-case delay and channel utilization than TDMA-based WSNs.

5.2 Ethical and Societal Considerations

Safety of humans and environment is an important ethical issue related to this work. Failing to satisfy the stringent timing and reliability requirements of safety- critical functions can result in dangerous consequences. Stringent regulations for health, safety, and the environment are now being enforced in many countries. These regulations require continuous monitoring of safety for workers at a plant so that help can be dispatched on time to prevent disastrous consequences that could endanger human lives, e.g., fire, explosion, leakage of poisonous gases. Moreover, security and privacy are legitimate concerns for IWSNs as wireless systems are vul- nerable to cyber threats. Attacks such as, denial-of-service, eavesdropping and radio interference can adversely affect the operation of the IWSN. Such attacks cause multiple disturbances to the network, e.g., loss of packets, jamming the channel, mod- ifying the network packets and/or inserting false packets into the network, which will eventually degrade the reliability of the network. Moreover, as a societal effect, the ubiquity of IWSNs technology everywhere in our surroundings has been a trigger for revolutionary applications that are beneficial for individuals, cooperations and governments. New jobs, industries, and economic models will emerge, and daily life will be changed in profound ways, allowing humans and machines to interact in unprecedented and unanticipated ways, or even replace the human workforce. This rise of workplace automation in its many forms has the potential to vastly improve productivity and augment the work of human employees. Automation technology can help remove the burden of repetitive administrative work and enable employees to focus on solving more complex issues while reducing the risk of error and allowing them to focus on value-added tasks. However, as a side effect, this could lead to displace workers from specific tasks that have been automated as a result of current and presumed technological advances. 5.3 Future Work 35

5.3 Future Work

Possible future investigations can be considered to handle the limitations associated with the proposed protocols. In PR-CCA, an efficient cross-layer scheduling method can be implemented to extend the work for multi-hop scenarios and evaluate the performance under the specifications of a real-world industrial applications, e.g., welder machine and plastic machinery. Furthermore, the effect of timing mis- alignment between the nodes should be considered, which will be more challenging for mesh networks, hence novel synchronization methods can be applied. For SS- MAC, the performance can be further investigated under realistic arrival models of the critical traffic. The proposed deadline-aware scheduling can be improved by integrating an efficient acceptance test that guarantees that all the scheduled transmissions meet their deadlines. In addition, novel routing algorithms can be integrated with SS-MAC in order to provide a deterministic delivery of the critical data in multi-hop scenarios. Moreover, recent promising technologies such as cognitive radio can be integrated with IWSNs to form an intelligent framework that can adapt the network performance in an efficient way to meet the strict requirements of time- and mission critical applications. 36 Bibliography

[1] J. Lu, H. Okada, T. Itoh, T. Harada, and R. Maeda, “Toward the world smallest wireless sensor nodes with ultralow power consumption,” IEEE Sensors Journal, vol. 14, no. 6, pp. 2035–2041, Jun. 2014.

[2] V. C. Gungor and F. C. Lambert, “A survey on communication networks for electric system automation,” Comput. Netw., vol. 50, no. 6, pp. 877–897, May 2006.

[3] P. Pleinevaux and J. D. Decotignie, “Time critical communication networks: ﬁeld buses,” IEEE Netw., vol. 2, no. 3, pp. 55–63, May 1988.

[4] U. Jecht, W. Stripf, and P. Wenzel, “Innovation and intellectual property rights,” in The Industrial Information Technology Handbook, R. Zurawski, Ed. Florida: CRC, 2005.

[5] L. Palopoli, R. Passerone, and T. Rizano, “Scalable ofﬂine optimization of industrial wireless sensor networks,” IEEE Trans. Ind. Informat., vol. 7, no. 2, pp. 328–339, May 2011.

[6] G. Cena and A. Valenzan, Operating principles and features of CAN. Florida: CRC, 2005.

[7] J. Blaney, “Wireless proves its value,” Power Engineering, 2009.

[8] M. Raza, N. Aslam, H. Le-Minh, S. Hussain, Y. Cao, and N. M. Khan, “A critical analysis of research potential, challenges, and future directives in industrial wireless sensor networks,” IEEE Commun. Surveys Tuts., vol. 20, no. 1, pp. 39–95, Firstquarter 2018.

[9] D. Chen, M. Nixon, and A. Mok, WirelessHART-Real Time Mesh Networks for In- dustrial Automation. New York: Springer, 2010.

[10] Q. Wang and J. Jiang, “Comparative examination on architecture and protocol of industrial wireless sensor network standards,” IEEE Commun. Surv. Tuts., vol. 18, no. 3, pp. 2197–2219, 3rd Quarter 2016.

[11] Z. Alliance, “Zigbee speciﬁcation,” Zigbee Document 053474r13, 2008.

37 38 BIBLIOGRAPHY

[12] S. Petersen and S. Carlsen, “WirelessHART Versus ISA100.11a: The Format War Hits the Factory Floor,” IEEE Ind. Electron. Mag., vol. 5, no. 4, pp. 23–34, Dec. 2011.

[13]J. Akerberg,˚ F. Reichenbach, M. Gidlund, and M. Bjorkman,¨ “Measurements on an industrial wirelessHART network supporting PROFIsafe: A case study,” in Proc. IEEE 16th Conf. Emerg. Technol. Factory Autom. (ETFA 2011), Sept. 2011, pp. 1–8.

[14] M. Wei and K. Kim, “Intrusion detection scheme using trafﬁc prediction for wireless industrial networks,” J. Commun. Netw., vol. 14, no. 3, pp. 310–318, Jun. 2012.

[15] IEEE Standard for Local and Metropolitan Area Networks-PART 15.4: Low-Rate Wire- less Personal Area Networks (LR-WPANs) Amendment 1: MAC Sublayer. IEEE Std 802.15.4e, 2012.

[16] R. T. Hermeto, A. Gallais, and F. Theoleyre, “Scheduling for IEEE 802.15.4- TSCH and slow channel hopping MAC in low power industrial wireless networks: A survey,” Comput. Commun., vol. 114, no. 13, pp. 84–105, Dec. 2017.

[17] L. M. Borges, F. J. Velez, and A. S. Lebres, “Survey on the characterization and classiﬁcation of wireless sensor network applications,” IEEE Commun. Surveys Tuts., vol. 16, no. 4, pp. 1860–1890, Nov. 2014.

[18] B. Glisiˇ c,´ “Performance and health monitoring of civil structures and infrastruc- ture using long-gauge and distributed ﬁber optic sensors,” in 2016 18th Interna- tional Conference on Transparent Optical Networks (ICTON), Jul. 2016, pp. 1–4.

[19] V. Sharma and S. Sharma, “Low energy consumption based patient health monitoring by leach protocol,” in 2017 International Conference on Inventive Systems and Control (ICISC), Jan. 2017, pp. 1–4.

[20] F. Tashtarian, A. Montazerolghaem, and M. Abbasi, “Distributed lifetime optimization of wireless sensor networks in smart grid,” in 2018 IEEE International Conference on Smart Energy Grid Engineering (SEGE), Aug. 2018, pp. 240–243.

[21] J. M. N. V., F. F. R., and Y. M. Q. L., “Design and implementation of wsn for preci- sion agriculture in white cabbage crops,” in 2017 IEEE XXIV International Con- ference on Electronics, Electrical Engineering and Computing (INTERCON), Aug. 2017, pp. 1–4.

[22] J. Aponte-Luis, J. A. Gomez-Gal´ an,´ F. Gomez-Bravo,´ M. Snchez-Raya, J. Alcina- Espigado, Teixido-Rovira, and P. Miguel, “An efﬁcient wireless sensor network for industrial monitoring and control,” Sensors, vol. 18, no. 1, 2018.

[23]J. Akerberg,˚ M. Gidlund, and M. Bjorkman,¨ “Future research challenges in wireless sensor and actuator networks targeting industrial automation,” in Proc. 9th IEEE Int. Conf. on Ind. Informat (INDIN 2011), Jul. 2011, pp. 410–415. BIBLIOGRAPHY 39

[24] P. Suriyachai, U. Roedig, and A. Scott, “A Survey of MAC Protocols for Mission- Critical Applications in Wireless Sensor Networks,” IEEE Commun. Surveys Tuts., vol. 14, no. 2, pp. 240–264, May 2012.

[25] F. Barac, M. Gidlund, and T. Zhang, “Scrutinizing bit- and symbol-errors of ieee 802.15.4 communication in industrial environments,” IEEE Trans. Instrum. Meas., vol. 63, no. 7, pp. 1783–1794, Jul. 2014.

[26] Y. Zou and G. Wang, “Intercept behavior analysis of industrial wireless sensor networks in the presence of eavesdropping attack,” IEEE Trans. Ind. Informat., vol. 12, no. 2, pp. 780–787, Apr. 2016.

[27] M. Bertocco, G. Gamba, A. Sona, and S. Vitturi, “Experimental characterization of wireless sensor networks for industrial applications,” IEEE Trans. on Instrum. and Meas., vol. 57, no. 8, pp. 1537–1546, Aug. 2008.

[28] C. M. G. Algora, V. A. Reguera, E. M. G. FernA¡ndez,˜ and K. Steenhaut, “Paral- lel rendezvous-based association for ieee 802.15.4 tsch networks,” IEEE Sensors Journal, vol. 18, no. 21, pp. 9005–9020, Nov. 2018.

[29] A. A. K. Somappa, K. O. vsthus, and L. M. Kristensen, “An industrial perspec- tive on wireless sensor networks: A survey of requirements, protocols, and challenges,” IEEE Communications Surveys Tutorials, vol. 16, no. 3, pp. 1391–1412, Third 2014.

[30] D. D. Guglielmo, S. Brienza, and G. Anastasi, “Ieee 802.15.4e: A survey,” J. Comput. Commun., vol. 88, pp. 1–24, Aug. 2016.

[31] P. Thubert, T. Watteyne, M. R. Palattella, X. Vilajosana, and Q. Wang, “IETF 6TSCH: Combining IPV6 connectivity with industrial performance,” in Proc. 7th Int. Conf. Innov. Mobile Internet Serv. Ubiq. Comput., Sept. 2013, pp. 541–546.

[32] H. Karl and A. Willig, Protocols and Architectures for Wireless Sensor Networks. JohnWiley & Sons Ltd, 2005.

[33] I. Rhee, A. Warrier, M. Aia, J. Min, and M. L. Sichitiu, “Z-MAC: A hybrid MAC for wireless sensor networks,” IEEE/ACM Trans. on Netw., vol. 16, no. 3, pp. 511–524, Jun. 2008.

[34] L. Ascorti, S. Savazzi, and S. Galimberti, “Cloud-based wirelessHART networking for critical industrial monitoring and control,” in Proc. 12th IEEE Int. Conf. on Ind. Informat. (INDIN 2014), Jul. 2014, pp. 362–369.

[35] M. Khanafer, M. Guennoun, and H. T. Mouftah, “A survey of beacon-enabled IEEE 802.15.4 MAC protocols in wireless sensor networks,” IEEE Commun. Sur- veys Tuts., vol. 16, no. 2, pp. 856–876, 2nd Quarter 2014.

[36] ”Wireless Medium Access Control and Physical Layer Speciﬁcation for Low Rate Wire- less Personal Area Networks”. IEEE 802.15.4 Standard, 2006. 40 BIBLIOGRAPHY

[37] J. Ploennigs, V. Vasyutynskyy, and K. Kabitzsch, “Comparative study of energy- efﬁcient sampling approaches for wireless control networks,” IEEE Trans. Ind. Informat., vol. 6, no. 3, pp. 416–424, Aug. 2010.

[38] J. Lin and M. A. Weitnauer, “Range extension cooperative mac to attack energy hole in duty-cycled multi-hop wsns,” Wireless Networks, vol. 24, no. 5, pp. 1419– 1437, Jul. 2018.

[39] K. Yu, M. Gidlund, J. Akerberg,˚ and M. Bjorkman,¨ “Performance evaluations and measurements of the REALFLOW routing protocol in wireless industrial networks,” IEEE Trans. Ind. Informat., vol. 13, no. 3, pp. 1410–1420, Jun. 2017.

[40] Y. K. Huang, A. C. Pang, and H. N. Hung, “An adaptive GTS allocation scheme for IEEE 802.15.4,” IEEE Trans. Parallel Distrib. Syst., vol. 19, no. 5, pp. 641–651, May 2008.

[41] A. Koubaa,ˆ M. Alves, E. Tovar, and A. Cunha, “An implicit GTS allocation mechanism in ieee 802.15.4 for time-sensitive wireless sensor networks: The- ory and practice,” Real-Time Syst., vol. 39, no. 1-3, pp. 169–204, Aug. 2008.

[42] S. e. Yoo, P. K. Chong, D. Kim, Y. Doh, M. L. Pham, E. Choi, and J. Huh, “Guar- anteeing real-time Services for industrial wireless sensor networks with IEEE 802.15.4,” IEEE Trans. on Ind. Electron., vol. 57, no. 11, pp. 3868–3876, Nov. 2010.

[43] L. Leonardi, G. Patti, F. Battaglia, and L. L. Bello, “Simulative assessments of the IEEE 802.15.4 CSMA/CA with priority channel access in structural health monitoring scenarios,” in Proc. IEEE 15th Int. Conf. on Ind. Informat. (INDIN 2017), jul. 2017, pp. 375–380.

[44] G. Patti and L. L. Bello, “A priority-aware multichannel adaptive framework for the IEEE 802.15.4e-LLDN,” IEEE Trans. on Ind. Electron., vol. 63, no. 10, pp. 6360–6370, Oct. 2016.

[45] M. Salajegheh, H. Soroush, and A. Kalis, “HyMAC: Hybrid TDMA/FDMA medium access control protocol for wireless sensor networks,” in Proc. IEEE 18th Int. Symp. on Personal, Indoor and Mobile Radio Commun. (PIMRC 2007), Sept. 2007, pp. 1–5.

[46] L. Sitanayah, C. J. Sreenan, and K. N. Brown, “A hybrid MAC protocol for emergency response wireless sensor networks,” Ad Hoc Netw., vol. 20, no. 9, pp. 77– 95, Sept. 2014.

[47] T. Zhang, T. Gong, C. Gu, H. Ji, S. Han, Q. Deng, and X. S. Hu, “Distributed dynamic packet scheduling for handling disturbances in real-time wireless networks,” in Proc. IEEE Real-Time and Embedded Technol. and Appl. Symp. (RTAS 2017), Apr. 2017, pp. 261–272.

[48] P. Suriyachai, J. Brown, and U. Roedig, “Time-critical data delivery in wireless sensor networks,” in Proc. 6th IEEE Int. Conf. Distrib. Comput. in Sensor Syst., 2010, pp. 216–229. BIBLIOGRAPHY 41

[49] B. Li, L. Nie, C. Wu, H. Gonzalez, and C. Lu, “Incorporating emergency alarms in reliable wireless process control,” in Proc. ACM/IEEE Int. Conf. Cyber-Phys. Syst. (ICCPS 2015), Apr. 2015, pp. 218–227. [50] X. Jin, C. Xia, H. Xu, L. Kong, and P. Zeng, “Mixed criticality Scheduling for industrial wireless sensor networks,” Sensors, vol. 16, no. 9, Aug. 2016.

[51] C. Xia, X. Jin, L. Kong, and P. Zeng, “Scheduling for emergency tasks in industrial wireless sensor networks,” Sensors, vol. 17, no. 7, Jul. 2017. [52] W. Shen, T. Zhang, F. Barac, and M. Gidlund, “PriorityMAC: A priority- enhanced MAC protocol for critical trafﬁc in industrial wireless sensor and actuator networks,” IEEE Trans. on Ind. Informat., vol. 10, no. 1, pp. 824–835, Feb. 2014. [53] T. Zheng, M. Gidlund, and J. Akerberg,˚ “WirArb: A new MAC protocol for time critical industrial wireless sensor network applications,” IEEE Sensors J., vol. 16, no. 7, pp. 2127–2139, Apr. 2016.

[54] L. Guntupalli, F. Y. Li, and J. Martinez-Bauset, “Event-triggered sleeping for synchronous dc mac in wsns: Mechanism and dtmc modeling,” in 2016 IEEE Global Communications Conference (GLOBECOM), Dec. 2016, pp. 1–6. [55] L. Guntupalli and F. Y. Li, “Dtmc modeling for performance evaluation of dw- mac in wireless sensor networks,” in 2016 IEEE Wireless Communications and Networking Conference, Apr. 2016, pp. 1–6. [56] J. Martinez-Bauset, L. Guntupalli, and F. Y. Li, “Performance analysis of synchronous duty-cycled mac protocols,” IEEE Wireless Communications Letters, vol. 4, no. 5, pp. 469–472, Oct. 2015. [57] I. Ramachandran, A. K. Das, and S. Roy, “Analysis of the contention access period of ieee 802.15.4 mac,” ACM Trans. Sen. Netw., vol. 3, no. 1, Mar. 2007. 42