Low-Complexity Design And Application In Digital Communication Systems

In modern communication systems,especially in the upcoming Internet of Things era,high speed,high stability,low power,low complexity and high security have become common communication system requirements.First,high speed and high stability mean errors and heavy Transmission should be avoided as much as possible.Low-density parity check(LDPC)code is used as standard code for data transmission channels in 5G NR standard.It is also widely used in various standards.Low power and low complexity mean the limitation of circuit scale.In this respect,approximate computing has emerged as a new circuit design method.In terms of security,physical unclonable function(PUF)has become a feasible option.This paper starts from the two major technical fields in modern communication systems,which are channel coding and secure communication,and completes the following four innovative tasks.First,we conduct a comprehensive analysis of the security of a new PUF,DC MUX PUF.We conduct special model analysis and attack experiments on it,including modeling and analyzing the functional logic of DC MUX PUF,deriving its functional functions.According to the function,we specially design its feature vector,and then we derive its partial derivative on each parameter and use the logistic regression algorithm to carry out repeated attack experiments on it.At the same time,we also use the same algorithm to attack MUX PUF.The average prediction rate of our attack experiment on MUX PUF is 91.17%to 91.04%(depending on the PUF order),and the highest prediction rate is 92.41%to 92.63%.The average prediction rate for DC MUX PUF is 88.61%to 89.01%,and the highest prediction rate is 89.89%to 90.77%.The average and highest prediction rates for DC MUX PUF are both lower.In terms of training time,the average time to attack DC MUX PUF is 1220.8 seconds.It is 40,000 times that of MUX PUF.Therefore it is found that DC MUX PUF has higher security in the face of model attacks than MUX PUF.Then in the field of channel coding,we complete the hardware design for the layered de-coding algorithm of LDPC code.Our decoder design is for the LDPC code used in the 802.11n(Wi Fi)protocol,and use a fully parallel structure.The results of synthesis and STA prove that the decoder design reaches a throughput rate of 2.275Gbps.The area is 2.296mm~2,and the area efficiency is 2.378Mbps/GE.Then we propose an approximate adder for 2's complement operation.It is suitable for LDPC decoders for its little effect on the decoding performance of the decoder.The delay gain compared to the precise decoder is from 19%to 23.55%(depending on the approximate order),the area gain is 12.60%to 50.39%,and the estimated power gain is from 1.91%to 6.46%.Finally,we propose a method for non-uniform deployment of approximate units in the LDPC decoder,and use this method to experiment with the scene of a 64-QAM cascaded LDPC fully parallel layered decoder.The decoding performance of the non-uniform approximate decoder is similar to that of OMS-C and OMS-C-AAU-3.At the same time,in terms of hardware perfor-mance,the throughput reaches 2.355Gbps,the area is reduced to 2.153mm~2.The area efficiency reaches 2.685Mbps/GE,which is an increase of 10.85%compared to the OMS-C version.In general,this article has completed the attack experiment against DC MUX PUF in the PUF field,completed its security analysis.Then the complete front-end design of the LDPC layered decoding algorithm is also finished.An approximate adder suitable for LDPC decoders is proposed.A method of non-uniform deployment of approximate units in LDPC decoders is proposed to help designers choose the placement and order of approximate units reasonably.