White Paper

Communications Systems Design

This white paper presents an over- view of system level design for communications equip- ment. It will be useful not just to those involved in the development of new communications systems, but also to designers of new prod- ucts using existing systems, or anybody seeking to develop new architectures and/or components to reduce product cost and/or size. It also describes the benefits and implications of developing highly Figure 1: Wireless street lighting control system (courtesy of Telensa) integrated (SoC) Introduction limitations, legislative restrictions, solutions, which dominate in today's (wireless) communications development budgets, unit cost targets high volume communications prod- systems have moved a long way in the and timescales. The design process is ucts. last 100 years. From radio broadcast- complex, requiring an in-depth under- ing, the first mass market application, standing of technology, radio systems, to today's ubiquitous digital processing and the ability to offering an array of applications and accurately simulate the effects of a communication at ever increasing data practical realisation at a system level. rates. This march of increasing func- tionality and complexity was enabled Applications and Requirements by the advent of the integrated circuit Although consumer demand for wire- and the microprocessor, which together less data is relentless, it would be have facilitated miniaturisation and wrong to think that all current develop- low-cost high volume manufacture. ments in communications are aimed at maximising traffic throughput over Modern communications systems pro- wide channel bandwidths. This type of liferate throughout the available spec- application certainly occupies an enor- trum and use a wide range of channel mous amount of global development bandwidths, schemes and effort, but there are also applications data rates. The main traffic for wireless where only a tiny amount of informa- communications systems is now data tion needs to be communicated between rather than voice and the required data sites and on an infrequent basis. rates are ever rising. When designing a communications system the imple- A good example of a modern low mentation choices depend on a host of data-rate application is in the monitor- requirements, some of which can be ing and control of street lighting, such conflicting. as that depicted in Figure 1 (courtesy of Telensa). These include the required functional- ity and performance, the available Here each street lamp is fitted with a

technology, operational conditions and small two way radio link allowing Sheet•Code••RFi0611

Plextek RF Integration Plextek Ltd., London Road, Great Chesterford, Essex CB10 1NY UK

T: +44 (0) 1799 533200 E: [email protected] W: www.plextekrfi.com monitoring and control by the local communications beyond the range of (excessive re-charging of mobile council (City authority). The council the normal cellular services. devices) but for other applications can control when the street are (vehicle security devices, remote meter turned on, and to save energy late into Satellite Satellite receivers are very reading devices) a battery life of many the night, when they can be dimmed to sensitive. The levels are very years is essential for effective commer- a lower level or switched off. The street low by the time they arrive on the cial deployment. Such a requirement can report back the state of its bulb ground and special attention is paid to must be considered from the outset and and ballast, or more recently LEDs, so minimising their figure. The will be a key parameter in the system that replacement can be undertaken down side is that they are more prone design process. when near end-of-life. The amount of to interference from high power data that needs to be communicated systems on nearby frequencies. Special Modulation Schemes and Data to/from each light is small and infre- protection is given to the satellite Rates quent. There is currently strong and frequency bands by restrictions on The one resource that all radio applica- growing interest in low data rate M2M other users nearby to ensure their tions have in common is the radio (machine-to-machine) communications radiated power was not excessive. spectrum and this has to be shared by such as this. However, the demand for growth of all users. As the number of applications cellular services has recently put this continue to increase, it becomes more Another growth area in communica- protection under pressure with the and more important that the use of this tions is high data rate links for personal allocation of cellular band frequencies resource is undertaken efficiently. consumer applications. High data-rate close to some satellite bands. WLAN and cellular communications is In all cases, the modulation must be now the norm and is expected by users. Battery life is an important design appropriate for the application, as effi- As higher and higher wireless data rates parameter in many communications cient as possible and constrained to the have become available, new applica- products. bounds of the allocated channel to tions evolve to make use of the addi- avoid spectral leakage into adjacent tional capacity. The volume of data that In some cases this simply needs to be must be transferred is greatly amplified long enough to avoid inconvenience in the back-haul network where the live data from thousands of users needs to be transferred between the local cellular tower and the cellular operator regional control centre.

Microwave line of sight links are often used for this purpose, operating at high frequencies where wide channel band- widths are available. In addition to using wide channel bandwidths, micro- wave links make use of higher order modulation schemes to squeeze more data through each channel.

Satellite systems allow truly global coverage. Global Navigation Satellite Systems (GNSS) provide one-way communications of navigational infor- mation worldwide. Two way satellite services are provided by other organi- sations and are used by emergency services in times of disasters as well as Figure 2: Combined I-Q constellation diagram by news reporters and those who need

Plextek RF Integration Plextek Ltd., London Road, Great Chesterford, Essex CB10 1NY UK

T: +44 (0) 1799 533200 E: [email protected] W: www.plextekrfi.com channels which could potentially inter- fere with users there.

There are many different modulation schemes operating in systems around the world today. For microwave point- to-point links, so called "high order" modulation schemes such as QAM64, QAM256 & QAM512 are commonly used. These offer high data transfer rates as each symbol of the modulation represents 6, 8 and 9 bits of data respec- tively. Their high data density comes at a price as it places challenging require- ments on the error vector magnitude (EVM), the recovered eye diagram and the control of the symbol detection points. Figure 3: BER versus Eb/No for different modulation schemes,M=4: To illustrate this, Figure 2 overlays four QPSK, M=16: QAM16 etc. separate modulation schemes on a single constellation diagram. The four distortion required to displace one par- The higher the order of modulation schemes: QPSK, QAM16, QAM32 & ticular constellation point from its used the higher the BER for a given QAM64 all have the same mean power region into that of another is smaller noise level. Figure 3 shows a plot of level, but it is clear that the individual than that required for a low order BER versus Eb/No for a range of constellation points of the highest scheme. modulation schemes, where Eb is the scheme (QAM64) are much closer to energy per bit and No the noise spectral each other than are those of the lowest The presence of noise in the receive density. It is clear that squeezing more scheme (QPSK). channel sets the fundamental limit on data through a given by the the Bit Error Rate (BER) that can be use of higher order modulation schemes This makes the higher order scheme achieved for a given modulation has an associated penalty in terms of much more susceptible to noise and scheme. the power that must be transmitted to distortion since the amount of noise or achieve the same BER (assuming a

Figure 4: Constellation and eye diagram for QAM256 modulation

Plextek RF Integration Plextek Ltd., London Road, Great Chesterford, Essex CB10 1NY UK

T: +44 (0) 1799 533200 E: [email protected] W: www.plextekrfi.com Figure 5: QAM 256 modulation spectrum (Blue) and Figure 6: QAM 256 modulation suffering transmit spectral mask (Green) for point-to-point links excessive spectral regrowth due to PA distortions

receiver with the same noise perform- In addition to corrupting the constella- It is possible to simulate the entire ance). There is also additional com- tion and causing eye closure, distortion communications system from bits in plexity in the radio design itself. in the transmitter can cause the signal to bits out. This is a time consuming to self-intermodulate and produce task but can be worthwhile particularly Figure 4 depicts the constellation and spectral regrowth (leakage) into adja- in cases where custom System on Chip eye diagram for QAM256 modulation cent channels. Figure 6 shows the (SoC) ICs are under development. without any real-world corruption. spectral regrowth that PA non-lineari- Figure 9 compares the simulated BER ties can cause. Here the regrowth to measured BER for a low-IF demod- In a practical system imperfections in exceeds the statutory limits imposed ulator designed by Plextek RFI and the transmitter, such as amplitude and by the regulatory authorities. implemented as part of a custom ASIC phase distortion in the RF now in volume production. The close and the phase noise of the local oscil- This level of distortion also corrupts agreement is a testament to the accu- lator would cause distortion, the eyes the transmitted constellation and eye racy that can be achieved with careful would close a little and the constella- diagram as depicted in Figure 7 and design and simulation. tion points would spread. Figure 8. Modern cellular systems make use of The frequency spectrum of a perfect Distortion within the receiver must Orthogonal Frequency Division QAM256 signal is depicted in Figure also be well contained. Channel filter- (OFDM) to further 5 together with the spectral mask ing, if applied too severely, can distort increase data throughput. OFDM can showing the emission limits for a the modulated waveform, resulting in be understood as a two-step process; microwave point-to-point link (green further eye closure and constellation first the data to be transmitted is split trace). Root-raised cosine filtering of distortion. This directly causes an into a large number of lower data-rate the transmitted data ensures its tight increase in bit error rate (BER). When streams. Each of these data streams is spectral control. It also introduces designing a radio system, sufficient then modulated onto a sub-carrier as Inter-Symbol-Interference (ISI) away time should be devoted to undertaking depicted in Figure 10, where each from the centre of the eye, as seen in system simulations to predict perform- subcarrier is orthogonal with the Figure 4. The receiver must maintain ance in terms of EVM, spectral others. Secondly, individual symbol timing with sufficient accuracy regrowth, constellation and eye distor- subcarriers are then modulated with to determine the symbol's value only tion and BER sensitivity. high order modulation to further at the eye opening point. increase capacity.

Plextek RF Integration Plextek Ltd., London Road, Great Chesterford, Essex CB10 1NY UK

T: +44 (0) 1799 533200 E: [email protected] W: www.plextekrfi.com Figure 7: QAM 256 constellation corrupted by PA non- Figure 8: QAM 256 eye diagram corrupted linearities by PA non-linearities The sub-carriers are orthogonal line-of-sight path between transmitting accounted for by modifying Equation because the carrier frequencies are and receiving antennas, the path loss 1 based on height and ground chosen so that there is no cross-talk follows a simple square law as defined reflection coefficient. For low angles between the sub-channels meaning in Equation 1, where d is the distance of reflection (long distance links) the guard bands between the sub-carriers between the two antennas, Gtx and Grx path loss tends to a 4th law model. are not required. The use of OFDM are the antenna gains and n =2. allows a high degree of compaction in Propagation and Link Budgets frequency giving a very high data This propagation model would be At microwave frequencies additional density per unit of transmission band- valid, for example, for satellite to loss is incurred due to rain. An allow- width. The advantage of OFDM over satellite communications where there ance for this (rain fade margin) must high data-rate single carrier schemes is one direct propagation path and no be made during the system design. The is its ability to cope with propagation additional paths due to reflections from magnitude of this allowance depends issues such as frequency selective objects. For ground based point-to- on the location of the link (and so the fading, this is discussed in more detail point links, reflections off the earth typical weather) and the target availa- below. mean that the free space model is no bility (percentage of time that the longer valid. This effect can be specified loss must not be exceeded Use of spatial diversity is also increas- ingly common in modern mobile devices.

This introduces two or more separate transmission paths between the and the cellular mast. These MIMO (multiple input multiple output) systems increase further the data throughput to and from the phone. Again, radio system design of these applications involves a lot of detailed simulation of the performance of MIMO OFDM modulated .

In free space, with a single direct Figure 9: Comparison of simulated to measured BER

Plextek RF Integration Plextek Ltd., London Road, Great Chesterford, Essex CB10 1NY UK

T: +44 (0) 1799 533200 E: [email protected] W: www.plextekrfi.com due to rain fall). Statistical data exists to allow appropriate rain fade margins OFDM Channel Bandwidth to be determined. OFDM There is also additional loss due to Subcarrier atmospheric absorption. This is both frequency dependent and altitude dependent but only really starts to become significant at mm-wave fre- quencies. The oxygen absorption peak at around 60GHz is an example of Frequency atmospheric absorption.

When designing a communications Figure 10: Simplified Image of OFDM system the link budget must be bal- anced to ensure that there is adequate gives rise to a large dip in the received The impact of this channel model on a signal power arriving at the receiver to signal strength. 5MBd QPSK signal having a 6.25MHz ensure acceptable Bit Error Rate bandwidth will now be illustrated. (BER) for the demodulated data. An approach to combat this is to imple- Note that the signal bandwidth will ment receive antenna diversity, where occupy the frequency range of The link budget can be improved by an additional receive antenna posi- ±3.125MHz about the zero Hz centre increasing transmit power, increasing tioned in a physically different location frequency. One would intuitively antenna gain (transmit and/or receive), is also available and will be subject to expect Response No.1 to have less reducing noise figure or changing different multi-path propagation impact on the signal when compared modulation scheme or rate so that the effects. to Response No. 2 because the ampli- required SNR at the receiver is tude distortion of Response No.1 over obtained. The impact of multipath propagation the ±3.125MHz band about zero Hz is on a transmission system can be much less than that for Response No. For non-line-of-sight propagation, assessed by undertaking a system sim- 2 and this is indeed the case. additional losses due to reflection, ulation with an appropriate multipath diffraction and scattering caused by model. A number of internationally Figure 12 depicts the eye diagram for obstructions must be included. The recognised models have been devel- the two responses. As expected, the simplest way to account for this is to oped to cover a range of different eye opening for Response No. 1 is increase the value of "n" in Equation propagation conditions and Figure 11 good and shows much less distortion 1 depending on the frequency, heights depicts the channel amplitude response than that for Response No. 2; the of the antennas and propagation envi- for two cases of the ITU Indoor Office position of the eye opening in the two ronment. For urban environments, Channel A model. The Indoor Office responses of Figure 12 has been cen- appropriate values of n can vary Channel A model is a 6-path static tred. between 2, when streets can act like model having an overall 35nsec rms waveguides, and 6, when significant delay spread. The corresponding constellation plot obstructions exist. for the two responses is presented in The difference between the two Figure 13 and again highlights the A more complex, but potentially more responses in Figure 11 is simply that better performance of Response 1. It representative, approach is to model Response No. 1 will occur at one RF shows that a channel having Response the propagation channel and include frequency and Response No. 2 will 2 will have more difficulties receiving this in the system simulation. It is occur at another RF frequency. Note the signal. Combatting the effects of possible for reflections to cause two that the x-axis is centred about 0Hz and multipath must be taken into account copies of the transmitted signal to this represents the centre frequency of when a system is being designed. arrive at the receiver with a phase the carrier. One effective means of combating ISI difference of close to 180°. This would is to periodically transmit a known be a specific case of flat fading and it sequence of data (a training sequence).

Plextek RF Integration Plextek Ltd., London Road, Great Chesterford, Essex CB10 1NY UK

T: +44 (0) 1799 533200 E: [email protected] W: www.plextekrfi.com Figure 11: Two instances of the propagation path amplitude response of ITU Indoor Office Channel A The receiver knows what it expects to a minimal variation. This means the receive signals from transmit signal so receive when this sequence is transmit- sub-carriers can now be recovered providing full duplex communications. ted and can remove the effects of ISI without the need for adaptive equalisa- These approaches all have different by adapting the weights of a digital tion of the channel, although it does benefits and which is most appropriate filter (an adaptive equalizer) to essen- require a more complex FFT based depends on the requirements of the tially create an inverse model of the demodulator. It should be noted that system. multipath interference caused by the whilst the need for equalisation may reflections in the channel. With have been removed, some sub-carriers Operating Frequency and moving users (and / or objects moving will be transmitted in propagation Interfering Signals in the local environment) the channel nulls, frequencies where there is sig- The radio frequency spectrum is a and its multi-path behaviour are con- nificant attenuation, and these subcar- global resource and its use is co-ordi- stantly changing so the filter will be riers may arrive at the receiver below nated on a worldwide basis by the continually adaptive. the noise floor and therefore not recov- International ered. The use of appropriate error Union (ITU); an agency of the United The use of OFDM, depicted in Figure correction and interleaving of data Nations based in Geneva. The ITU 10, also helps to combat ISI. In this across sub-carriers can be used in publishes a table of frequency alloca- case the high data rate signal is split mitigation against this effect. tions that represents agreement into multiple (n) sub-channels of a amongst the 191 member states of the lower data rate. Each of these sub- Multiple Access Techniques ITU. It separates the globe into three channels is separated in frequency by To allow multiple users access to the radio regions on geographical grounds: the reciprocal of the symbol rate to same communications system, a means Ÿ minimise interference between other of separating the different signals from Region 1: Europe, Middle East, sub-channels. The resulting signal each other must be used. This can be Africa,the former Soviet Union, comprises n sub-carriers each with a by separation in time (Time Division including Siberia; and Mongolia; Ÿ data rate and modulation bandwidth Multiple Access - TDMA), in fre- Region 2: North and South reduced by a factor n. This allows each quency (Frequency Division Multiple America and Pacific (East of the sub-channel to propagate independ- Access - FDMA) or by applying a International Date Line); Ÿ ently along a multipath channel with unique code (Code Division Multiple Region 3: Asia, Australia and the very little amplitude and phase varia- Access - CDMA). Pacific Rim (West of the Interna- tion for that reduced bandwidth. tional Date Line). There will still be amplitude variation Combinations of these techniques are Frequency allocations may differ to between sub-channels, but for an indi- also used. Separation in time and/or some extent between the three regions vidual sub-channel the aim is to have frequency is also used to separate and the table is revised periodically at

Plextek RF Integration Plextek Ltd., London Road, Great Chesterford, Essex CB10 1NY UK

T: +44 (0) 1799 533200 E: [email protected] W: www.plextekrfi.com Figure 12: Eye diagram for 5MBd QPSK signal subjected to ITU Indoor Office Channel A multipath model. Response No.1 (LHS) and Response No.2 (RHS)

World Radiocommunications Confer- Any communications system must operating at other frequencies. This ences. Individual countries are respon- adhere to the appropriate legislation requires care and attention during the sible for managing the radio spectrum and the choice of operating frequency design process. The main interference in their own jurisdiction and generally has a huge impact on the cost, achiev- mechanisms experienced in a receiver follow the lead given by the ITU table, able performance and design complex- are: but they are free to modify the table to ity. Ÿ Insufficient selectivity where suit their own needs. Any modification imperfections in a receiver allow will usually be undertaken with the aim Radio spectrum is a shared resource, an adjacent transmission to corrupt of not causing interference to other so other users could be operating on a wanted signal. This can be com- countries as these will be assumed to nearby frequencies. Whatever the bated by additional filtering within be abiding by the ITU plan. operating frequency of the system the receiver. there will be strict limits on the Ÿ Reciprocal mixing - the phase Each country controls access to use the allowed emissions at all other noise sidebands of the local oscil- frequency spectrum in their jurisdic- frequencies and this can be a lator are sufficiently high to cause tion by a variety of means. For some significant design challenge, an unwanted signal to be down- parts of the spectrum, the country particularly for spectrum close to the converted into the receive band regulator will set regulatory limits and operating band. Problems that can give Ÿ Insufficient intermodulation issue licenses to control the interfer- rise to unwanted emissions from a immunity - receiver imperfections ence between users; an example of this transmitter include: allow transmissions from two or being users of private mobile radio more signals to interact within the systems. For other parts, the regulator Ÿ Harmonic emissions receiver and generate an on-chan- provides less restriction and relies on Ÿ Non-harmonic spurious emissions nel product that corrupts the users to self-control their emissions; Ÿ Adjacent channel leakage of wanted signal …. even though the an example of this would be cellular carrier phase noise interfering transmissions may be operators. Yet other parts of the spec- Ÿ Spectral regrowth in adjacent well separated from the wanted trum are designated licence-exempt channels caused by carrier self- channel and here there is relatively weak regu- intermodulation Ÿ Spurious responses - unintended latory control of interference levels; an Ÿ Spurious transient emissions mixing products translate an example of this is the Industrial, Sci- caused by transient effects unwanted signal into the receive entific and Medical (ISM) bands. band It is also important that communication Ÿ Blocking - where a very large receivers can operate in the presence signal overloads a receiver of legitimate communications devices

Plextek RF Integration Plextek Ltd., London Road, Great Chesterford, Essex CB10 1NY UK

T: +44 (0) 1799 533200 E: [email protected] W: www.plextekrfi.com Figure 13: Constellation diagram for 5MBd QPSK signal subjected to ITU Indoor Office Channel A multipath model. Response No.1 (LHS) and Response No.2 (RHS)

attempting to receive a weak architectures to make use of the advan- wanted signal tages of integration whilst steering protection against blocking signals and around the disadvantages. For instance reject transmitter harmonics Double Careful transceiver design is required it is possible to realise integrated ana- conversion designs are used in appli- to ensure the problems listed above are logue filters having useful selectivity cations requiring significant IF gain. minimised for the application. Appro- at low frequencies but not at RF fre- The predominance of discrete filtering priate filtering must be included, the quencies, so receiver architectures makes the architecture unsuited for individual building blocks must have have been developed or adapted to integration onto silicon and so unable adequate linearity, LO and IF frequen- make use of this. Such architectures to fully benefit from the miniaturisa- cies must be selected with care, local include low-IF and zero-IF types and tion it can offer. This was the main oscillators must have adequate phase are the preferred approaches for highly reason that new architectures were noise performance, Automatic Gain integrated . sought, which were better suited in Control (AGC) may be necessary. integration. Figure 14 shows a block diagram of a Radio Architectures single conversion heterodyne trans- Figure 15 shows a block diagram of a A broad range of radio architectures ceiver; an architecture that goes back zero-IF transceiver. This was one of have been developed. For high volume to the early days of radio when designs the first developed with full silicon world wide applications, an aim will were implemented using discrete com- integration in mind. The received always be to seek the size and cost ponents. The receiver mixer converts signal is mixed directly down to base- benefits of an integrated solution. A the signal from the transmission fre- band where low-pass filters now Bluetooth transceiver is one such quency to a standard IF such as provide the channel filtering. Low pass example where the whole transceiver 10.7MHz, 21.4MHz, 45MHz or analogue filters at baseband are suited architecture is realised in silicon. 70MHz where crystal or ceramic filters for integration and can be tailored to a Lower volume applications may be provide narrow band channel filtering desired response. Two baseband chan- realised by a mixture of integrated and with high selectivity. Similar filtering nels (I and Q) are required to preserve discrete components. in the transmitter ensures the transmit- both amplitude and phase information ted signal only broadcasts on the in the received signal. A drawback with The move to full silicon integration wanted channel. Filtering at RF is used the architecture is that it is susceptible over the last twenty or so years has to reject image responses caused by the to self-reception of its own local oscil- been achieved by adapting transceiver mixing process, provide lator with the result that a DC compo-

Plextek RF Integration Plextek Ltd., London Road, Great Chesterford, Essex CB10 1NY UK

T: +44 (0) 1799 533200 E: [email protected] W: www.plextekrfi.com Figure 14: Single conversion heterodyne architecture

nent (a DC offset) appears in the the received RF signal is mixed down Matching in the low-IF strip and quad- baseband alongside the wanted signal. to a very low frequency IF - close to rature is essential to minimise the DC - which, as with the zero-IF leakage of the local oscillator and This DC component needs to be approach, allows the channel filters to leakage of the image signal to the removed and this can be done by either be fully integrated. output, however the degree of match- DC blocking with a low cut-off high ing needed to meet regulatory require- pass filter or by a feedback arrange- These “polyphase” filters make use of ments is often insufficient and an ment that measures the DC offset and the quadrature IF channel and produce external bandpass filter may be cancels it by adding an equal but a passband response at the frequency required. opposite DC component via a DAC. of the low-IF and a rejected response Another disadvantage to this architec- at its image frequency. The amount of This approach can be problematic and ture is that it is susceptible to 2nd order image rejection achieved by this archi- so some solutions will use the simpler distortion whereby the envelope of an tecture is dependent on the degree of zero-IF architecture for the transmitter unwanted interferer is translated matching that can be achieved in the and adopt a mixed low-IF receive, directly to baseband. mixers and IF and the quality of the zero-IF transmit design. quadrature LO signal driving the The zero-IF transmitter is well suited mixers. The degree of image rejection Further simplification of an integrated to integration with the baseband mod- will be finite and is likely to limit the architecture may be possible when the ulated signal appearing immediately at receiver selectivity at this frequency. operation of the radio system is consid- final frequency. As before, two identi- An advantage to the low-IF receiver ered. For example: cal baseband channels are required to compared to the zero-IF (direct conver- correctly produce the output modu- sion) is that the self-reception that Ÿ A TDD system has only the trans- lated waveform. Imperfections in the produced a problematic DC compo- mitter or the receiver operating at mixer will give rise to leakage of the nent can be overcome by the addition any one time and this makes it local oscillator signal appearing on the of a DC block. A second down conver- possible to share one local oscilla- output, and to combat this, a DC offset sion stage in the digital domain can be tor between the two parts. correction (cancellation) would need added to obtain a baseband signal Ÿ Some protocols may have periods to be applied. output. where it is know that there will be Figure 16 shows a block diagram of a A low-IF transmitter converts a quad- no signal to receive, such dead low-IF transceiver; an alternative rature modulated signal at a low IF design for full integration. In this case directly up to the final frequency.

Plextek RF Integration Plextek Ltd., London Road, Great Chesterford, Essex CB10 1NY UK

T: +44 (0) 1799 533200 E: [email protected] W: www.plextekrfi.com Figure 15: Direct (zero-IF) conversion architecture

periods can be used to undertake Ÿ The trade-off between sensitivity frequency and applied to an ADC for self-calibration routines such as: and linearity will require careful digitisation and then subsequently specification of the individual filtered, amplified and detected in the o Cancelling a DC offset in the blocks that make up a receiver digital domain. The adaptability of receive IF strip, chain. Too much receiver gain, such a software implementation pro- o Tuning a banded VCO to the whilst favouring sensitivity, will vides a high level of flexibility to the correct frequency, also reduce dynamic range and application areas. o Minimising the local oscilla- loss of performance when in the tor feed-though in a transmit- presence of strong interfering sig- System Simulation ter, nals. Noise figure is often the Designing and implementing a new o Minimising the image guiding benchmark for the ulti- communications system is time con- response by adjustment of the mate sensitivity that can be suming and expensive. Effective sim- local oscillator I-Q phase and obtained whilst intercept point (for ulation of the whole communications amplitude balance both second and third order distor- system prior to implementation is tion) is a benchmark for high level essential. For the RF chain a cascaded As in all engineering, trade-offs will interference immunity. spread-sheet analysis of parameters form part of the design process, for such as NF, gain, IP3, IP2 and com- example: Ÿ The transmitted EVM and adjacent pression was the traditional approach channel leakage power will and this is still a valuable tool during Ÿ The receiver selectivity that can be depend on the quality of the local the development process. achieved will depend on a range oscillator, the modulation format of factors including the degree of and complexity, the linearity of the Various system simulator packages are filtering implemented, the quality transmit strip amplifiers and their now commercially available that allow of the local oscillator, specifically associated noise floors, etc. this approach to be taken a step further its sideband noise and for non- with the incorporation of aspects such direct conversion architectures, the Some, but not all, applications will be as filter responses, phase noise profiles amount of image rejection that can able to take advantage of a software of local oscillators and the effects of be achieved. defined radio approach - where the out of band interferers. received signal is mixed to a low

Plextek RF Integration Plextek Ltd., London Road, Great Chesterford, Essex CB10 1NY UK

T: +44 (0) 1799 533200 E: [email protected] W: www.plextekrfi.com Figure 16: Low-IF architecture

To fully optimise the performance of are highly recommended. a communications system it is neces- sary to undertake a complete end to Concluding Remarks end communications system analysis The development of a new communi- from bits in to bits out, including the cations system is a huge undertaking. effects of the channel itself. This In many cases the use of an existing requires co-simulation of digital and system and/or standard is the most analogue circuitry and is no small sensible approach offering shortest undertaking. For practical reasons timescales to deployment, lowest systems are often developed with a development costs and lowest risk. slightly scaled back approach, where However, when new application sce- sections of the system are simulated narios or new functional requirements independently. arise and the expected volumes justify the development costs a custom com- If a highly integrated custom IC is to munications system can lead to per- be developed, either for a new system formance and cost benefits. This white or as a cost/size reduction exercise for paper has given an overview of the an existing system, the need to ensure communications system design adequate performance through simula- process and discussed implementation tion is even more vital. The cost of choices and development require- committing to an integrated implemen- ments. tation is high, the options for imple- menting changes after fabrication are restricted and the timescales and costs of a second iteration cause nightmares for many project managers. In this instance simulation of the complete communications system, such as the BER simulation included in Figure 9,

Plextek RF Integration Plextek Ltd., London Road, Great Chesterford, Essex CB10 1NY UK

T: +44 (0) 1799 533200 E: [email protected] W: www.plextekrfi.com