Performance Comparison of Wireless Speech Communications Schemes

Exper iments a r e conducted using a number of adaptive transmission systems that a r e suitable for th i rd generation personal communications networks (PCN) . I t becomes apparent that the selection of the t y p e of modem and number of modulation levels of the modem, the speech codec t o b e used, and whether t o deploy a channel codec, depends on the channel S N R a n d teletrafflc demand if the power consumption of the hand held portable is t o b e minimised. Following the implications of our experiments, we generalise o u r findings t o some of the requirements of a n adaptive transceiver that might b e deployed in the next generation of mobile radio networks. 1. I N T R O D U C T I O N There is much activity world wide in attempting to define the third generation personal communications network (PCN). In designing a third generation system cognizance is given to the second generation systems. Indeed, we may anticipate that some of the sub-systems of GSM and DECT may find their way into PCNs either as a primary sub-system, or as a component to achieve backward compatibility with systems in the field. This approach may result in hand-held transceivers that are intelligent multimode terminals, able to communicate with existing networks, while having more advanced and adaptive features that we would expect to see in the next generation of PCNs. In this text we present the results of a series of experiments we conducted. We concerned ourselves with the radio link and introduced a menu that includes some second generation sub-systems for backward compatibility. We also included 16level quadrature amplitude modulation (16QAM), as we consider that this type of modulation can be made adaptive and reconfigure itself, for example as ;-shifted differential quadrature phase shift keying (5-DQPSK) and Gaussian minimum shift keying (GMSK). Essentially our deliberations centre around the concept of adaptive multichannel coding that may be required. In this short study we restricted ourselves to the microcellular propagation environment. 2. M O D E M S C H E M E S The choice of modem is based on the interplay of equipment complexity, power consumption, spectral efficiency, robustness against channel errors, cochannel and adjacent channel interference, as well as the propagation phenomena which depends on the cell size [I]. Equally important are the associated issues of linear or non-linear amplification and filtering, the applicability of non-coherent, differential detection, soft-decision detection, equalisation and so forth. In our experiments we used three different modems, namely GMSK, S-DQPSK and 16StQAM, each with a lowand a high-complexity detector. In our GMSK modem a normalised bandwidth bit-timing product of BT = 0.3 was favoured, which was adopted by the Pan-European second generation system known as GSM. The typical bandwidth efficiency of GMSK is about 1.35 bit/Hz. A low-complexity frequency discriminator and a higher-complexity Viterbi detector were deployed [I]. S-DQPSK modulation was favoured by the Americans in their IS-54 D-AMPS network and by the Japanese digital cellular system. We decided to compare the performance of a low-complexity differential detector to a more complex maximum likelihood correlation receiver (MLH-CR). We used square root raised cosine Nyquist filters with a roll-off factor of 0.35 and achieved an approximate modem bandwidth efficiency of 1.64 bit/Hz. The differentially coded 16level twin-ring star QAM ( 1 6 StQAM) included in our studies was shown to work well , in microcellular environments, where high SNRs and low dispersion are generally the norm and the high traffic density requirements justify the use of linear amplification. The low-complexity non-coherent differential detector's performance [2] was compared to that of a MLH-CR. When using square root raised cosine Nyquist filters with a roll-off factor of 1 the modem bandwidth efficiency was 2.4 bit/Hz. level modulation and the types of appropriate speech and 3. S P E E C H C O D E C S MOMUC-2, 2ND INTERNATIONAL WORKSHOP ON MOBILE MULTIMEDIA COMMUNICATIONS, 11-13 APR., In recent years both speech coding research and IC-technolog! 1995, BRISTOL, UK @IEE went through a revolutionary development, culminating in a number of standards with bit rates as low as 4.8 kbps [I]. The selection of the speech codec for mobile applications is based on the appropriate combination of parameters such as speech quality, computational complexity related to power consumption, bit rate, delay and robustness against channel errors. The two speech codecs included in our experiments are the CCITT G 721 32 kbps adaptive differential pulse code modulation (ADPCM) codec used in the DECT system, and the 13 kbps GSM and DCS 1800 Regular Pulse Excited (RPE), Long Term Predictor (LTP) assisted codec. The ADPCM codec used in our schemes fully complies with the CCITT Recommendation so we will not describe its opera-