Research On Receiver Signal Detection Technology For Indoor High-speed MIMO-OFDM System

In recent years,with the needs of people's lives and social development,traditional wireless communication technology has been difficult to meet people's application needs.The new generation of application requirements,such as ultra-high definition video,virtual reality(VR),etc,are constantly emerging.Therefore,the wireless short-range transmission technology with higher throughput,higher reliability and lower delay has become a new research hotspot.At present,the widely used multiple input multiple output orthogonal frequency division multiplexing(MIMO-OFDM)technology puts forward higher requirements for the signal detection technology of MIMO receiver when solving the problem of large capacity data transmission.In the face of these difficulties and challenges,this paper makes a deep research on the receiver signal detection technology of indoor high-speed MIMO-OFDM system from the two aspects of basic algorithm improvement and VLSI architecture design.The main research contents are as follows.First of all,this paper introduces the background and significance of the research and the research status at home and abroad,and summarizes the difficulties faced by MIMO receiver multi antenna signal detection technology.Based on this,the typical indoor channel characteristics and system architecture of MIMO-OFDM receiver signal detection system are presented.In particular,the paper analyzes the key and difficult problems faced by the signal detection technology of high throughput,high reliability and low delay.Subsequently,for the problem of large-capacity and low-latency short-distance transmission under complex indoor conditions,this paper studies the channel preprocessing and tree search algorithm in the real-domain K-Best detection algorithm that approximates ML,including: Firstly,in view of the complexity of real domain channel preprocessing,it is proposed to adjust the sequence of column vectors in the channel matrix pair with column vector 1l-norm as reference,so that the layer with high signal-to-noise ratio can be detected first,so as to achieve better detection performance.Furthermore,the square root sign operation and unit division operation in traditional QR decomposition are converted into coordinate rotation digital computer(CORDIC)to reduce the complexity of operation.Finally,aiming at the problem that the number of layers of the real domain K-Best search algorithm is doubled,according to the characteristics of the R matrix after the improved channel matrix preprocessing,we propose to expand the adjacent two layers of sub nodes at the same time,so as to reduce the detection delay in the real domain detection system.Finally,for the hardware implementation of the proposed channel preprocessing algorithm and K-Best search algorithm,the optimization of each key link and VLSI multiplexing architecture are proposed.Compared with the traditional VLSI implementation of channel matrix preprocessing,this scheme can save 18.6% multiplier resources by using 1l-norm to sort the vectors in pairs.At the same time,the CORDIC unit is used to calculate the module and the unit of the pair of column vectors,so as to avoid the root opening and division operations in the traditional methods.Furthermore,look-up table(LUT)is used to expand parallel sub nodes in the K-Best search phase,which reduces the k-1 clock cycle consumption compared with the traditional node enumeration method.In addition,by optimizing the parity parallel sorting unit,the comparator number can be reduced by 50% compared with the traditional Z-type sorting rule,and the comparator number can be reduced by 22.8% compared with the traditional parity parallel sorting unit.In summary,compared with the traditional implementation,the VLSI solution proposed in this paper has more advantages in the delay and complexity of MIMO signal detection.Theoretical analysis and simulation experiments show that the indoor high-speed MIMO-OFDM system can achieve a throughput rate of nearly 2Gbps while approaching Maximum Likelihood(ML)detection performance and reducing detection delay.