.gsm / Gprs System Physical Layer Algorithm Simulation And Zsp Core-based Optimization

GSM is the most widely used communication system today. This theis researchs on the algorithms in the physical layer of the GSM system and test its implementation on 'YYY', a multi-mod SoC chip which is designed by our country. All the research work here is based on GSM 05 series specifications released by 3GPP.Firstly, we will introduce the research background of this theis, including YYY chip and the importance of our software simulation to the chip. Then we outline the GSM radio subsystem.In the second chapter, the GSM modulation system is introduced. Firstly, we give the flow chart of the transmitter and receiver of the whole system and introduce briefly about channel encoding, interleaver and burst building in the transmitter. Secondly, we present the GMSK modulation that GSM uses, including its principles and implementation in a digital storage way.The third chapter is about the synchronization. Due to the Rayleigh eclipse and time delay caused by the wireless channel, the primary task of the receiver is the synchronization work. It is aimed to eliminate all the errors so that the signals can be correctly demodulated. In this chapter, we introduce the frame structures in GSM firstly and then describe the realization of the frequency synchronization by FCCH and time synchronization by SCH.Channel estimation is explained in chapter 4. Firstly, we list out the multi-channel models used in GSM and simulate them by the modified JAKES model. Then we focus on the channel estimation, including the channel delay estimation by the auto-correlation of the training sequence, the adaptive channel order estimation by noise floor method and the channel impulse estimation by LS method. All the parameters we get here will be used in the equalization process.The fifth is the key chapter in this thesis. It is about the adaptive equalization in GSM system, which is used to eliminate the inter symbol interfere and demodulate the signals. We introduce MVA firstly and then compare two kinds of soft-decision algorithms (SOVA and VALPP) and do some further improvement on VALPP. We also add the frequency estimation and compensation in the normal burst. The frequency estimation algorithm used here makes use of the hard decision bits in equalization process and is the modified realization of decision feedback estimation.The sixth chapter is about viterbi decoding and its realization on ZSP 540 core. We give the performace of the core for viterbi algorithm and compare it with other cores. The last chapter is the analysis of the system's performace, including VALPP vs hard decision in the TU, RA, HT channels and the BER rate of the system in the three channels. The results show that it is match with the demand specified in GSM specifications.