Research On Receiving Algorithms Of LTE-A Uplink Data Channel Based On General Purpose Processor

As a rapid development global wireless communication standard in recent years, the LTE/LTE-A continuously moves forward which is driven by the requirements of improving system capacity, users’high QoS and high transmission rate. Meanwhile, its various key techniques and the realtime processing are also one of the important research directions of technical personnel. Considering the performance, power con-sumption and various aspects, General Purpose Processor (GPP), a competitive gen-eral processing chip, can not only realize the realtime of wireless signal processing system, but also serve as a valuable base and reference for this paper in building the simulation platform.During the evolution of LTE and LTE-A, the antenna size keeps increasing which leads to higher receiving algorithm complexity. In the receiving algorithms of Uplink Data Channel, MLD (Maximum Likelihood Decoding) has the best perfor-mance and the biggest complexity at the same time. An efficient way to reduce the algorithm complexity is to narrow the search range from the whole detection space to a Multidimensional super ball, which is called Sphere Decoding as well. Under LTE-A uplink4*4antenna system, this paper takes simulations on different MIMO detecting algorithms include ZF(Zero Forcing), MMSE(Minimum Mean Square Error) and SD(Sphere Decoding) to get the BER(Bit-Error-Ratio) curve based on GPP. Fu-thermore, an improved initial function of SD search radius based on existing research is proposed in this paper, which makes a further decrease as1.76of algorithm processing latency in the low SNR range.However, with the continuously increasing of antenna size, the realization com-plexity of data receiving algorithm are also rising. How to optimize data process in simulation system and reduce process time delay is not only an important direction of this paper, but also a supplementary verification for the improvement of signal real-time receiving process. Therefore, another main research content of this paper is to make use of SIMD set, multi-thread and multi-core parallel method to optimize si-mulation procedures, in the premise of guaranteeing module algorithm performance. From the delay results before and after the optimization, we can draw the conclusion that the decrease of module delay is proportional to the algorithm parallel degree. The conclusion shows that the use of GPP has great potential, at the same time predict that realize the wireless signal processing based on GPP in the base station may be the main development direction of future signal processing system.