【摘要】：如今無線通信飛速發(fā)展,傳統(tǒng)的ASIC實(shí)現(xiàn)方式已經(jīng)難以滿足當(dāng)前對移動終端更新?lián)Q代的需求,因此基于軟件無線電的實(shí)現(xiàn)方式越來越受到大家的歡迎,目前各大公司都在研究基于處理器無線通信的SDR實(shí)現(xiàn)。本文主要研究無線通信基于多核處理器的實(shí)現(xiàn)方式,本文選取了移動廣播多媒體CMMB和4G標(biāo)準(zhǔn)LTE兩個(gè)典型標(biāo)準(zhǔn)。 本文主要工作主要包括： 1)分析當(dāng)前國內(nèi)外的無線通信標(biāo)準(zhǔn),并且調(diào)研了當(dāng)前國內(nèi)外各大研究單位針對SDR實(shí)現(xiàn)方式的主流處理器平臺構(gòu)架,并通過分析典型的處理器平臺的優(yōu)缺點(diǎn),分析歸納出適用于通信處理器平臺的一般優(yōu)化方法。 2)針對CMMB內(nèi)接收機(jī)部分,由于其算法已經(jīng)比較成熟,因此我們采用現(xiàn)有典型的算法。從算法層面,我們選擇了較優(yōu)的算法劃分,根據(jù)數(shù)據(jù)流進(jìn)行任務(wù)劃分,盡量細(xì)化每個(gè)任務(wù),開發(fā)其中的并行性。映射過程中根據(jù)增大并行性和減少通信量兩大主要準(zhǔn)則進(jìn)行合理的映射到多核處理器平臺。接著從SIMD方面增大并行度,寄存器擴(kuò)展,共享存儲器利用以及優(yōu)化硬件添加復(fù)雜指令對實(shí)現(xiàn)進(jìn)行優(yōu)化。最終整個(gè)內(nèi)接收機(jī)的吞吐率能達(dá)到120Mbps,完全滿足CMMB需求。與當(dāng)前比較先進(jìn)的實(shí)現(xiàn)結(jié)果進(jìn)行對比,本文實(shí)現(xiàn)結(jié)果在吞吐率上有絕對優(yōu)勢,并且能效比僅為2.19nJ/bit低于當(dāng)前其他實(shí)現(xiàn)結(jié)果。 3)針對LTE的MIMO檢測,本文首先分析了對于MIMO檢測的各種算法(線性和非線性的檢測算法),然后著重分析了當(dāng)前最熱門的球形譯碼算法中的FD-BF和K-Best算法,在FD-BF在的基礎(chǔ)上,提出了一種基于K-Best的枚舉策略的KE-FDBF算法。該算法在保證誤符號率的情況下,其訪問節(jié)點(diǎn)的個(gè)數(shù)遠(yuǎn)遠(yuǎn)小于K-Best以及訪問節(jié)點(diǎn)不定的FD-BF算法。實(shí)現(xiàn)時(shí)本文結(jié)合了第二版本的多核處理器的新特性,并且通過堆棧策略的優(yōu)化,DMA以及添加一些不易用處理器實(shí)現(xiàn)的Demap和Cordic處理單元,大大提升了處理效率。其實(shí)現(xiàn)結(jié)果與當(dāng)前ASIC實(shí)現(xiàn)結(jié)果較為接近。CMMB內(nèi)接收機(jī)和LTE的MIMO信號檢測良好的實(shí)現(xiàn)結(jié)果表明了我們映射方式的合理以及我們的多核處理器平臺在通信領(lǐng)域有著巨大的潛力。
[Abstract]:With the rapid development of wireless communication, the traditional ASIC implementation has been difficult to meet the current needs of mobile terminal updating, so the implementation based on software radio is more and more popular. At present, all the major companies are studying the implementation of SDR based on processor wireless communication. This paper mainly studies the implementation of wireless communication based on multi-core processor. Two typical standards of mobile broadcast multimedia CMMB and 4G standard LTE are selected in this paper. The main work of this paper is as follows: 1) analyze the current wireless communication standards at home and abroad, and investigate the mainstream processor platform architecture for the implementation of SDR among the major domestic and foreign research units. Through the analysis of the advantages and disadvantages of the typical processor platform, the general optimization method for communication processor platform is summarized. 2) for the part of CMMB receiver, because the algorithm is mature, we adopt the typical algorithm. From the algorithm level, we choose the better algorithm partition, divide the tasks according to the data stream, refine each task as far as possible, and develop the parallelism among them. In the process of mapping, it is reasonable to map to multi-core processor platform according to the two main criteria of increasing parallelism and reducing traffic. Then from the aspect of SIMD to increase parallelism, register expansion, shared memory utilization and optimization hardware to add complex instructions to optimize the implementation. Finally, the throughput of the whole receiver can reach 120 Mbps, which fully meets the requirements of CMMB. Compared with the current advanced implementation results, the proposed implementation results have absolute advantage in throughput, and the energy-efficiency ratio is only lower than other current implementation results in 2.19nJ/bit. 3) for the MIMO detection of LTE, this paper first analyzes various algorithms for MIMO detection (linear and non-linear detection algorithms), and then analyzes the FD-BF and K-Best algorithms in the most popular spherical decoding algorithms. On the basis of FD-BF, a KE-FDBF algorithm based on K-Best enumeration strategy is proposed. When the symbol error rate is guaranteed, the number of access nodes in this algorithm is much smaller than that of K-Best and the FD-BF algorithm with uncertain access nodes. This paper combines the new features of the second version of the multi-core processor, and through the optimization of stack strategy, DMA and some Demap and Cordic processing units which are not easy to implement with the processor, the processing efficiency is greatly improved. The implementation results are close to those of the current ASIC implementation, and the good results of MIMO signal detection in CMMB receiver and LTE show that our mapping mode is reasonable and our multi-core processor platform has great potential in the field of communication.
【學(xué)位授予單位】：復(fù)旦大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2012
【分類號】：TN92;TP332

【參考文獻(xiàn)】

相關(guān)博士學(xué)位論文 前1條

1 吳川;數(shù)字電視解調(diào)芯片關(guān)鍵技術(shù)研究[D];復(fù)旦大學(xué);2011年

，

本文編號：2438789