Soft Base Station Based On HIHG Performance General Purpose Processor And ITS Resource Allocation Reserach

Soft base station, based on General Purpose Processor (GPP), can make a full use of GPP in its flexibility and scalability, to support multiple wireless communication protocols. Thus, soft base station is considered to be one of key technologies in5G systems. To face the challenges introduced by soft base stations, this dissertation combines the Computer Science and Wireless Communications, to solve the problems of low real-time, low accuracy and low efficiency in soft base stations. Then, based on the research, a prototype of3GPP LTE is established and tested. The main work of this dissertation is listed as follows:1) In this dissertation, the features of baseband digital signal streams are analysed as well as the data structure. A parallel optimization architecture is proposed, and realizes decoupling of data streams. Then, the algorithm is considered to be a multi-objective optimization problem. It tries to reduce the space complexity while maintaining the real-time demand. Theoretical analysis and system verification results show that the proposed optimization methods can obtain a gain of10-630.2) In OFDM systems, synchronization schemes based on training sequence have a relatively low complexity. However, the performance is poor. In this dissertation, the structure of training sequence is redefining to reduce the relevance between cyclic prefix and data symbols. And a low searching dimension synchronization scheme is proposed. Furthermore, based on the mapping method in soft based station, a synchronization error estimation and compensation scheme is proposed to enhance the performance. Simulation results show that the sytem can obtatin a BER gain of9dB.3) The architecture of future RAN (Radio Access Network) is evolving towards centralized baseband. However, a great number of mobile users will introduce a intensive computing demand. Thus, a CSI (Channel State Information) based computing resource allocation scheme is proposed:firstly modeling the resources; then obtaining relationship between channel decoding iteration times and resource demand; finally deducing the expression of computing resources demand as a function of CSI. Further, a computing resource adaptive allocation scheme is proposed. Simulation results show that the system throughput can obtain a gain of1dB.4) Based on previous research, the general architecture of soft base station, as well as clock synchronization scheme is proposed. Then, a3GPP LTE soft base station prototype is eastablished based on a Intel i7CPU. Through the transmission test, the GPP utilization as well processing rate is obtained. It shows that the LTE soft based station can reach a peak data rate of50.12Mbps, as well as a BER (Bit Erro Rate) of1.37×10-6. What’s more, its clock jitter is within a level of μs, which outperforms most of existing soft base stations.This dissertation carries out the research on GPP-based soft base station and its resource scheduling from above four respect. It solves the problem of real-time demand when apply wireless baseband algorithm into GPP, and also explores the adaptive computing resource scheduling of centralized baseband in future wireless network architecture. Currently, the reseach on5G system is started in full swing, and GPP-based soft base station becomes a key point in5G technologies, which is adopted in National863Plan Projects.