`
`WIRELESS OFFICE DATA COMMUNICATIONSUSING CT2 AND DECT
`
`J J Spicer, G A Halls! and G Crisp?
`Introduction
`
`Advances in personal communications technology have been accompanied by parallel computer
`developments. Technology synergy has shown the technicalfeasibility to develop wireless data
`applications, and both CT2 and DECT provide support for office data communications.
`In this paper
`we describe the services supported and address some implementationissues.
`
`These new generationsof cordless telephones were primarily designed to suppor digitally encoded
`voice and employdigital transmission techniques in order to improve both the spectral efficiency and
`the grade of service. This same radio radio system architecture can also be usedto carry data traffic;
`there are however some important implementation considerations that must be addressed before such
`services can be supported.
`
`The CT2 standard, [1], is being enhanced to support circuit-mode data-bearer services, which would
`potentially allow cordless data links in the office for a range of low rate (up to 19.2kbits/s) services,
`eg E-mail, file printing, smal! file transfer, etc. The DECT standard, [2], can support multiple
`32kbits/s data channels, offering a capability for higher rate services, thereby facilitating applications
`requiring larger volumes of data exchange. Systems based on these existing cordless standards,
`Tather than newer spread spectrum technology, would offer economic benefits arising from the
`economies of scale expected from the high volume cordless telephone market, as well as, in the
`longer term,facilitating integrated voice and data servicesin the office environment, [3].
`
`Radio Architecture
`
`both CT2 and DECT were designed with the primary aim of providing cordless telephonyservices.
`Although the potential for carrying data was recognised this waS a secondary issue in the
`developmentof the system. This means that the organisation of the frame structures are specifically
`tailored to a form required to carry digitised speech; this does little actual harm to the potential for
`carrying data but it does result in frame formats which need careful use if data is to be handled
`successfully.
`
`The CT2 radio interface uses frequency division multiplexing, FDM, to provide 40 channels
`(i)
`at 864.1MHz to 868.1MHz. Timedivision duplex, TDD,is used to give a bi-directional cordless
`link. Gaussian Frequency Shift Keying, GFSK, at 72kbps is used with bursts of 72 bits, Ims,
`transmitted and received alternately. Once the call has been established properly, 64 of these bits are
`available for traffic, normally speech but in our case data, giving an unprotected data rate of 32kpsin
`each direction.
`
`The DECT radiointerface operates at a carrier frequencyin the region of 1.88GHzto 1.9GHz
`Gi)
`using GFSK at 1.152Mb/s. Time division multiple access, TDMA, provides timeslots for up to 12
`users, with TDD giving the duplexing. A TDMAframeis 10mslong and contains 24 slots, 12 for
`the base station to handset, downlink, and 12 for the uplink: each slot carmies 320bits of information,
`giving a data rate of 32kb/s. For data applications a single link is allowed to use more than oneslot
`in each directionto increase the user data rate, asymmetric assignmentsare also permitted.
`
`13 J Spicer and GA Halls are with Roke Manor Research Lid.
`
`2 G Crisp is with GPT Lid.
`
`8/1
`
`Authorizedlicensed uselimited to: Fish & Richardson PC. Downloaded on April 28,2022 at 15:14:28 UTC from IEEE Xplore. Restrictions apply.
`
`1
`
`SAMSUNG 1018
`
`1
`
`SAMSUNG 1018
`
`

`

`
`
`Radio
`
`Cl
`
`LBehavi
`
`The channel characteristics enjoyed by the radio system will dictate the performance that can be
`delivered to the users. There are three essential elements which characterises the channel:
`propagation, interference and noise. Thestatistics of these are used to select the types and levels of
`error control to be applied.
`In the classical treatmentof error control, steady channel statistics are
`assumed andthe error control can be finely tuned to give optimum performance.
`In our present case
`however, the propagation and interference can vary wildly with time and the error control regime
`must be carefully designed to cope. First of all then, we must determine the essential channels
`statistics.
`
`CT2 uses a carrier frequency in the region of 866MHz,using a narrowband modulation. For
`(i)
`indoor, wireless office applications the noise and interference are expected to remain fairly constant,
`but the received signal envelope is expected to have a Rician probability density function with
`dynamic range 30dB and fading bandwidth of ~4Hz. These temporal variations are due to movement
`in the vicinity of the cordless radio link causing shadowing and frequency selective fading.
`Multipath and inter-symbolinterferenceeffects are expected to be negligible for this application since
`the bit period is much greater than the maximum measured delay path. The CT2 errorstatistics are
`therefore expected to be bursty in nature: within a lms transmission frame, the bit error probability
`will change very little and can be regarded as a constant. For the most part the error probability will
`be very low and slowly changing, depending mainly on the range between transmitter and receiver,
`but occasional signal fades will occur causing error probabilities of up to 50%.
`
`(ii)|DECT operates at a carrier frequency around 1.9GHz, using a wideband modulation. As
`with CT2, the noise and interference are expected to remain fairly constant, the propagation
`characteristics however are likely to be quite different. The measured delay spreads are reported as
`being only of the order of tens of nanoseconds, consequently the intersymbol interference is
`dominated by frequency selective fading. This generates narrowband notches in the wideband
`modulation causing the demodulatorto create effectively random errors as the nodulation frequency
`sweepsacross a notch. The temporal nature of the errorstatistics is again governed by movement,
`with infrequent error bursts of short duration. The DECT channel can be expected to provide a
`normally low background bit error probability, 1 in 104 to 1 in 105 with rare, short bursts of errors.
`
`
`
`EnorControl
`
`As mentioned earlier, both CT2 and DECT were designed to carry speech traffic and as such do not
`provide any in built error protection. This is quite adequate for speech where an occasional error
`would be indiscernible to the user, and a short burst of errors would manifestitself as a click and not
`disrupt any conversation too greatly. Such errors on a data channel would, in general, be disastrous:
`an averageerror rate of 1 in 105 bits would be totally unacceptable, and this is far better than an
`unprotected radio channel could be expected to provide. So some form of error control must be
`applied to give a user error rate better than 1 in 10!°, if any data service is to gain a useful place in the
`market.
`
`There are two basic ways of ensuring a low error rate to the user. Both require the addition of
`redundant information to the users information to generate a message which can be decodedat the
`receiver, [4].
`In the first, the decoder detects the presence of errors in the message and requests a
`repeat, ARQ; in the second the decoderis able to both detect and correct the errors (forward error
`correction FEC) without the need for a repeated transmission. In general terms, ARQ will give a low
`undetected error rate at the cost of reduced throughput whilst FEC will maintain throughputat the
`cost of an increased undetectederrorrate.
`
`Anyrealistic system must provide a balanced service to the user and consequently a combination of
`ARQ and FECis often chosen. Another importantfactor is the amount of redundancy required to
`provide the degree of protection required by the user: ARQ needsless redundancy than FEC in order
`to provide the same undetected error rate. But when errors are detected, a re-transmission is required
`
`9/2
`
`Authorizedlicensed uselimited to: Fish & Richardson PC. Downloaded on April 28,2022 at 15:14:28 UTC from IEEE Xplore. Restrictions apply.
`2
`
`2
`
`

`

`
`
`which reduces the channel throughput. For a medium errorrate channel, which stays fairly constant.
`il may prove more efficient to use FEC, which despite having an initially lower throughput achieves a
`higher overall throughput because of the absence of re-transmissions.
`
`As discussed above, the CT2 channel delivers a wide range oferror probabilities which are
`GQ)
`fairly slowly changing. The scheme designed to cope with this channel uses both error correction
`and ARQ for the asynchronous services and FEC alone for the synchronous services. Reed
`Solomon block codes are used to provide the redundancy and a procedure, LAPR, (based on CCITT
`LAPB) controls the flows of data in the asynchronous mode. The codes used are based on 8 bit
`symbols and are shortened to be 63 symbols long. A frame is made up from one codeword and a
`single 8 bit synchronising symbol: thus it takes 8 bursts lasting 16ms to wansmit. The amount of
`information contained in each codeword depends on the service being supported and the rate
`selected, varying from 2 symbols to 44 symbols:
`this information must contain the controlfield for
`the procedure as well as the user data.
`
`The Reed Solomon codes are a powerful set of block codes which are able to detect and correct
`errors on a symbol by symbol basis.
`(A single bit error in a symbol or all eight bits in error are
`classed as a single symbolerror.) The code is used to correct a small numberoferrors, butif this
`number is exceeded, a repeat request is issued by LAPR. Simulations and experimentation have
`shownthat this protocol is able to support the specified data rates at extremely low error rates, | in
`10!2 wascalculated for the asynchronousservice, [5].
`
`The DECTchannel changes quite rapidly and covers a medium rangeof error probabilities.
`(i)
`A scheme to cope with this uses ARQ alone, accepting the occasional losses and retransmissions
`with the compensation of a simpler implementation. This latter point 1s of great importance when
`multiple slots are used. Depending on the protocol being used, one of two fixed length packets
`containing information, control field and checksum are sent. The checksum is used purely for error
`detection; retransmissions are requested when errors are detected. Simulations have shownthatthis
`schemeis able to support very high rate data transmission at acceptably low errorrates.
`[6] shows
`that rates up to 400xbits/s can be supported for single users and rates around 8O0kbits/s can be
`achieved for multiple users.
`
`Services
`
`Cordless telephony provides the business user with the freedom to roam around his organisation’s
`premises with the ability to both receive and make telephone calls.
`In the same way cordless
`technology can provide similar facilities for data communications. With the increasing use and
`decreasing size of computers, the business useris likely to become more dependent on data
`communications and expect the same mobility as for telephony applications. Other scenarios include
`the interconnection of less portable devices such as workstations in anefficient and convenient way;
`cordless technology offers a simple low cost way of providing these connections.
`
`full-duplex asynchronous and full-duplex
`CT2 offers two main classes of service:
`qa)
`synchronous.
`In thefirst of these, an asynchronous packet assembler/disassembler, PAD, provides
`a CCITT V24/V28 (RS232) device to the user.
`In the context of the wireless office this could be
`used to provide a wireless connection, for example between a terminal and a computeror between a
`workstation and a printer. The standard data rates specified for this service are 300, 1200, 2400,
`4800, 9600, 14400 and 19200bits/s; these will be suitable for small file transfers and electronic mail
`type applications, but quite impractical for large file wansfers. The synchronousservice offers the
`samestandard data rates and in addition an unprotected 32kbit/s mode. This service could be used to
`provide a wireless connectioninto an ISDN for example.
`
`(ti)|The DECT protocols include options for interworking with local area networks, LAN. This
`makestheir application in the wireless office very appealing. Where either small file transfers, or a
`wireless connection to a host computer via a LAN are required, the data rates needed can be
`supported using a single time slot in each direction. Where high data rates are needed,for example to
`transfer large files, then multiple slots can be used to increase the datarates.
`
`9/3
`
`Authorized licensed uselimited to: Fish & Richardson PC. Downloaded onApril 28,2022 at 15:14:28 UTC from IEEE Xplore. Restrictions apply.
`
`3
`
`3
`
`

`

`
`
`Conclusion
`
`The inclusion of data communications services into the CT2 and DECT standardsoffers the facility to
`provide wireless office capabilities in cost effective manner. The CT2 standard offers a range of
`asynchronous and synchronousservices. The DECT standard offers the possibility of higher data
`tates and includes access to local area networks.
`
`References
`
`[1]
`
`{2]
`
`(3)
`
`(4)
`
`[5]
`
`{6}
`
`I-ETS 300 131.
`
`ETS 300 175series.
`
`“Cordless Personal Communications”, WHW Tuttlebee, EEE Communications
`Magazine, Vol 30, pp42-53, Dec 1992.
`
`“Error Control Coding”, S Lin and DJ Costello, Prentice-Hall, Englewood Cliffs, 1983.
`
`“A Protocol for the Transmission of Data over CT2”, IM Davis, Roke Manor Research
`Report, 72/9 1/R/075/U, March 1992.
`
`“The Application of the Digital European Cordless Telecommunications (DECT) Standard to
`Local Area Networks (LANS)”, GA Halls and GC Davis, IEE Colloquium on Radio
`LANS, May 1992.
`
`9/4
`
`Authorized licensed uselimited to: Fish & Richardson PC. Downloaded onApril 28,2022 at 15:14:28 UTC from IEEE Xplore. Restrictions apply.
`
` 4
`
`4
`
`