随着智能时代的到来，移动通信的发展也迈进了新的台阶。因此，科学家们在第四代（fourth generation, 4G）移动通信技术的基础上提出了对第五代（fifth generation, 5G）移动通信技术的设想。除了传输速率有质的飞跃外，5G最大的改进之处是能灵活地支持各种不同的智能设备，如健身跟踪器和智能手表等，这对5G提出了极大的挑战。大规模多输入多输出（multiple input multiple output, MIMO）技术是5G的关键技术之一，可以支持数以百计的用户进行传输，而信号检测器是大规模MIMO系统中的重要一环，其主要作用是从接收信号和已知信息中恢复和分离出发送信号，对整个系统的性能其到重要作用，因此，对信号检测算法进行研究变得极为关键。论文首先对大规模MIMO系统中几种经典的线性和非线性信号检测算法分别进行了简单的介绍。线性信号检测算法包括迫零（zero forcing, ZF）信号检测算法、最小均方误差（minimum mean square error, MMSE）信号检测算法、纽曼级数近似（Neumann series approximation, NSA）算法和高斯-赛德尔（Gauss-Seidel, GS）算法，非线性信号检测算法包括最优的最大似然（maximum likelihood, ML）信号检测算法、K-Best信号检测算法和球形译码（sphere decoding, SD）信号检测算法。然后论文提出了一种新的信号检测算法?——改进并行迭代（modified parallelizable iteration, MPI）信号检测算法，该算法以迭代的方式来近似矩阵求逆，同时采用新的求解初值的方法来代替传统的零初值。仿真显示MPI信号检测算法在降低计算复杂度的同时还能实现接近于MMSE信号检测算法的检测精度。接着对大规模MIMO系统中可重构信号检测处理器的架构进行设计，包括对数据通路和配置通路的设计。最后给出了该可重构信号检测处理器的实验结果，验证可得该处理器可以支持多种信号检测算法和常用的数字信号处理领域的算法，而且在功耗、性能和能效方面均优于数字信号处理器（digital signal processor, DSP）。

With the improvement of living standards, the demand for wireless communication increases accordingly. Based on fourth generation mobile communication system, scientists put forward ideas about fifth generation (5G) mobile communication system. As one of key technologies of 5G, massive MIMO technology can support hundreds of users to transmit. The signal detector is one critical part of massive MIMO systems, whose function is to reveal and separate transmitting signals from receiving signals and other known information. Because the signal detector plays an important role in system performance, the research for signal detection algorithms of high detection accuracy and low computational complexity is crucial. Meanwhile, how to design the flexible hardware architecture who keeps low power and high performance is also challenging.This paper first simply introduces several linear and nonlinear signal detection algorithms in massive MIMO systems. Then Modified parallelizable iteration (MPI) algorithm is proposed. MPI algorithm approximately calculate matrix inversion in an iterative way. Meanwhile, a new initial solution is adopted to instead of traditional zero initial solution. Simulation shows that MPI algorithm can approximately achieve the performance of MMSE algorithm and decease the computational complexity. The architecture of reconfigurable signal detection processor in massive MIMO systems is designed next. In the end, this paper shows the test results of reconfigurable signal detection processor, which is better than FPGA in area efficiency and energy efficiency. Moreover, the processor can support various signal detection algorithms and typical digital signal processing algorithms.This paper design the hardware architecture of reconfigurable signal detection processor for MPI algorithm. First, using Verilog HDL to describe the circuits and obtain the simulation results. Then, using TSMC 28nm technology to synthesize the circuit. At last, obtaining the circuit layout.