MODELING AND PERFORMANCE EVALUATION OF MOBILE VHF AS MODULATION SCHEME WIO CHANNELS EMPLOYING FSK, DPSK, QPSK AND 8-ary PSK
F. Swarts, Hendrik Christoffel Ferreira · 2000
In this paper we present the results of experiments undertaken to find suitable models for mobile V.H.F. channels. The experiments were carried out using four different modulation schemes at different signalling rates. The modulation schemes used were: FSK @ 300 baud, DPSK @ 1200 baud, QPSK @ 1200 baud, and 8 - ary PSK @ 1600 baud. The experiments were undertaken for urban as well as freeway driving. Fritchman partitioned Markov chain models were applied throughout and from the models, block error probability distributions were derived. These block error probability distributions or P(m,n), give the probability that a block of n transmitted bits will contain exactly m errors. In this paper we present P(m,n) for 7, 15 and 31 bit blocks, for the abovementioned modulation schemes, on a mobile V.H.F. channel. The channel models presented in this paper should be of value when simulations of communication systems employing the abovementioned modulation schemes have to be undertaken. Furthermore the P(m, n) information presented, should give some indication of the performance to be expected from block error correcting schemes. I INTRODUCTION The importance of mobile digital communication is ever on the increase (2). It is therefore important to determine the performance that can be expected from mobile digital communication systems. With this paper it is our intention to present channel models that reliably represent mobile digital communication systems. Our study involves the use of four modulation schemes namely FSK, DPSK, QPSK and 8 - ary PSK, for the transmission of digital data from a vehicle moving (i) through a densely populated urban area and (ii) along a freeway, to a stationary receiving station. At the receiving station the received data is compared to the transmitted data, errors identified and statistics pertaining to the nature of the occurrence of errors recorded. From the recorded statistics, discrete Markov chain models representing the statistical distribution of the errors, can be found. These models would then allow studies of the channel to be undertaken without the need of having the channel available. Furthermore these models can also be applied in the assessment of the performance of various coding schemes. We shall also briefly present the methodology for determining the models as well as the P(m,n) distributions. Mabey (9), and French and Mabey (lo) undertook similar studies to ascertain values for P(m,n) in the U.H.F. band. Their work did however not include a study of channel models or various modulation schemes as we have done here. Their investigations did look at two data rates namely 1200 bits/s and 4800 bits/s. Finally, some of the most important conclusions to be drawn from the results will be presented. II CHANNEL MODELS AND BLOCK ERROR PROBABILITY DISTRTBUTIONS Before starting the discussion of the channel model itself we shall briefly discuss some of the concepts used to describe error statistics. To briefly present the relevant concepts for describing an error sequence, consider the following extract from an error sequence. 1's denote error bits, 0's denote correct bits. d denotesx successive correct bits: . .I 0l9O 1 o2 1 o7 1 1 o~~~ 1 0~~~1 o~~~ 1 1 I. . v--- GAP BURST BURST-INTERVAL CLUSTER