Research On Key Technologies Of Multi-core Processor

In recent years, along with the innovation of science and technology, communication technology, which is closely related to people’s daily lives, is constantly developing, therefore the corresponding communication standard protocol is rapidly upgrading. The traditional ASIC-based digital baseband implementations have encountered great challenges in terms of R&D update rate and the flexibility of the system configuration. On another hand, in the research field of processor, single-core processors have reached bottleneck on improving the performance along with technology development, and using multi-core processors to improve processor performance has gradually become the major solution. Especially for mobile communications applications, the multi-core processors have high flexibility, which includes shorter development cycles, more compatible and more flexible on communication standard application programming, so that multi-core processor-based platform can effectively solve ASIC’s problems in realizing the mobile communication baseband. It can be foreseen that applications based on multicore platform in the field of communications must have bright prospects.This paper studies the LTE downlink system critical module’s (channel decoding) algorithm and mapping method on multi-core processors, analyzes the advantages and disadvantages of multi-core processors in applying the communications applications, and then study the ways to enhance the performance of communications application based on the multi-core processors.The main work of this dissertation include:1)This dissertation gives a brief analysis of the radio channel characteristics and the non-ideal factors that affect the performance of the communication system. It introduces the origin of channel code theory, the development of Turbo code, and the significance of adoption of Turbo code on channel transmission to enhance the performance of the error correction process.2) It gives a comparative analysis of the advantages and disadvantages between the different algorithms used in the Turbo decoder, including MAP algorithm, Log-MAP algorithm and MAX-Log-MAP algorithm. Through the simulation and complexity analysis of these different kinds of algorithms related to performance, combined with the structural characteristics of multicore processor platforms, this paper finally adopted MAX-Log-MAP decoding algorithm, which gives a effectively balance between the complexity and performance of the decoding operations, and is suitable for Turbo decoding realization for practical applications3) Turbo decoder tasks are re-devised and integrated, in order to fully exploit the intrinsic link between the sub-tasks, including the correlation data between parallel or sequential tasks, and then are mapped on the multicore processor platforms. Take full advantage of the acceleration unit in the multi-core platform structure, as well as the shared memory and the global chip network communication structure, the performance of Turbo decoding is maximized.4) The paper gives a quantitative analysis on the simulation data of Turbo decoder’s mapping scheme on multi-core processor. Make the amount of task load allocated to each core balanced as far as possible, to avoid idle waiting states that will affect decoding performance. Meanwhile, due to the implementation of a pipeline structure, the uniform distribution of tasks is beneficial to minimize the critical path length of the pipeline and to enhance the performance of the pipeline. Through reasonable task scheduling, reducing the proportion of inter-core communications times to the overall running time, the effective operation times of the processor is increased.