Font Size: a A A

Research On Finite Length High Reliability Analog Fountain Codes

Posted on:2020-02-28Degree:MasterType:Thesis
Country:ChinaCandidate:Z X HuangFull Text:PDF
GTID:2428330590474100Subject:Information and Communication Engineering
Abstract/Summary:PDF Full Text Request
The fifth generation(5G)mobile communication will support ultra-reliable low latency communication(URLLC)services to meet the ultra-low latency and high reliability wireless communication needs of emerging I nternet of Things(IoT)services in the future,such as automatic driving,tele-surgery,smart grid,etc.These services require a communication delay of less than 1 millisecond while satisfying a block error rate(BLER)of less than 10-5.Since low latency is equivalent to using a short blocklength code in channel coding,this will result in a reduction in coding gain inevitably.Therefore,the physical layer coding technology for URLLC is a key problem in 5G technology,which has triggered an extensively research and attention recently.The latest short blocklength performance bound provides theoretical guidance for finite blocklength high performance channel coding design.On the one hand,although the existing URLLC coding scheme can approach the short blocklength performance bound,most of them have the disadvantages of fixed rate and high complexity.On the other hand,the time-varying characteristics of the wireless channel make our URLLC channel coding required the rateless characteristic,adaptively “online” encoding according to the channel state,in order to avoid the delay caused by retransmission.Based on this,we introduce a M-ary rateless code with linear complexity—Analog Fountain Code(AFC),and optimizes for the URLLC requirements under the short blocklength cases.The specific research contents are as follows:Although the infinite blocklength AFC can approach the Shannon limits in a wide range of signal-to-noise ratios(SNR),there is still some problem in the optimization of finite blocklength AFC.Note that for the rateless codes,average output nodes degree is the key to the improving of the decoding performance.Therefore,the second chapter first analyzes trend of BLER perf ormance between AFC degree and blocklength under finite blocklength cases.Secondly,the concatenated code is a classic scheme to improve the performance of finite blocklength code,so we designed two concatenate schemes of low density parity check(LDPC)codes and AFC.The simulation results verifies that the above two LDPC-AFC concatenate schemes can effectively improve the BLER performance,but far from the short blocklength performance bound.In the case of short blocklength,merely increasing the average of AFC will cause a large number of short loops in the encoding tanner graph,and lead to the appearing of error floor.In the third chapter,we introduce the Progressive Edge Growth(PEG)algorithm to reconstruct the edge of the encod ing tanner graph for short blocklength AFC.The proposed coding algorithm can minimize the occurrence probability of short loops and improve the BLER performance of finite blocklength AFC.Furthermore,since the short loops in short blocklength AFC is inevitable,we make the weight optimization(WO)by combine the weight coefficient characteristics of AFC,and proposed the WOPEG-AFC encoding algorithm,assign different sides for different weight coefficients,further enhances the BLER performance.The simulation result verif ies that both PEG-AFC and WOPEG-AFC can approach the short blocklength performance bound at high SNR.Based on the above work,in order to further improve the performance of finite block length AFC,firstly we derive the performance limit of finite block length AFC in AWGN channel in the fourth chapter.Theoretical and simulation results show that in the high SNR interval(SNR?10dB),the performance of finite block length AFC will increase with the increase of weight set information entropy;While in the low SNR interval(SNR<10dB),the performance of finite block length AFC depends mainly on the code rate.Therefore,for diffe rent SNR intervals,we design the optimal weight sets which following the Gaussian distribution.And then,based on the finite feedback mechanism,we proposed the weight adaption-Analog Fountain Code(WA-AFC)transmission scheme,so that the finite block length AFC can reach the performance limits with the lowest complex in a wide range of SNRs.
Keywords/Search Tags:ultra-reliable low latency communication, finite length code, analog fountain code, weight distribution, complexity
PDF Full Text Request
Related items