Design And Optimization Of Key Technology In2G/3G Physical Layer Based On General Purpose Processor

With a rapid development of wireless communication technology in recent years, traditional radio access network communications system has been unable to meet the current needs of business growth. In response to increasing requirements of quality of service, the transmission rate and system capacity, C-RAN as a new wireless access network architecture emerged. Before coming of the mature4G network, as mobile communication systems widely used currently,2G and3G are important members of C-RAN. On the other hand, considering the performance, power consumption and various aspects, as a competitive general processing chip, General Purpose Processor (GPP) is an addition option for baseband signal processing except the dedicated signal processor.Based on such a technical background, this paper build a simulation platform for GSM and TD-SCDMA system based on GPPs, and test the bit error rate(BER) performace and delay performace of the simulation link. This paper focuses on design and optimization of GSM physical layer and TD-SCDMA joint detection module based on GPPs. This paper gives a detailed key technologies introduction of GSM physical layer and their implementations. After analyzing ZF-BLE and MMSE-BLE algorithm, we selects ZF-BLE as the appropriate joint detection algorithm of TD-SCDMA system on GPP platform, and then discusses optimizing ZF-BLE algorithm using approximate cholesky decomposition. This paper lists three optimization methods based on features of general purpose processor, including lookup table, SIMD parallel instructions and IPP libraries, and then discusses concrete realizations of these optimization methods when optimizing those modules.The main innovations of this paper are the following two points:First, compare DSP with GPP, and summarize three optimization direction namely the lookup table, SIMD parallel instructions, IPP library based on characteristics of GPP platform; second, propose rapid look-up table method for CRC, convolution coding, GMSK modulation of GSM and channel stimulation response vector calculation of joint detection, and giving complete table and lookup method, propose two SIMD parallel optimization methods to speed up complex multiplications in the receiver of GSM and joint detection of TD-SCDMA and propose merging operation to use IPP library and reduce loop consumption.This paper also simulates the BER performace of channels and delay perfermace of each modul. Seen from results of simulation, GSM and TD-SCDMA system which implemented and optimized on GPPs meet requirements of the BER and real-time processing. It verifies the signal processing capability of GPPs. Compared with dedicated signal processors, GPPs can undertake multi-tasks, have a low price, better system compatibility and code portability. Beside, with using nmerous optimization methods, the real-time signal processing capability based on GPPs will be further improved. It shows eormous potential of undertake the signal processing task of future wireless communication systems.