Research On Sequential Scheduling Strategies For LDPC Decoding And RCM Demodulation

The reliability and validity of information transmission are two important performance indicators of communication systems.Error-correcting coding and rate adaptation are effective methods to improve the reliability and validity of information transmission,respectively.Low-density parity-check(LDPC)code is an error-correcting code with Shannon limit approaching decoding performance and low computational complexity.Rate compatible modulation(RCM)is a method that can achieve smooth rate adaptation on time-varying channel.The standard belief propagation(BP)decoding algorithm for LDPC codes is a parallel algorithm in which the updating of messages takes the form of flooding,that is,in each round of iteration,all check nodes are updated first,subsequently all variable nodes are updated.Since the latest message generated by the update process cannot be fully utilized in the current iteration,the convergence rate of the standard BP decoding is slow.Sequential scheduling strategies are effective methods to accelerate the decoding convergence of LDPC codes.They can be categorised into standard sequential scheduling(SSS)strategies and informed dynamic scheduling(IDS)strategies.To some extent,each of the IDS strategies is affected by the "greediness" problem inherited from the residual belief propagation(RBP)algorithm,which hinders the further improvement of the decoding performance.In order to alleviate the greediness of RBP algorithm,we propose an improved algorithm based on RBP decoding algorithm,called Damped-RBP decoding algorithm,which introduces a damping factor in the calculation of residuals.The effect is that if a message is repeatedly assigned the highest priority of update,the residual of this message is scaled down from the original calculated value,the multiple of scaling down increases exponentially with the number of repetitions.Simulation results show that the proposed algorithm can effectively alleviate the greediness of RBP decoding algorithm.The message updates in the existing RCM demodulation algorithms take the form of flooding,and the convergence rate of demodulation is slow.Inspired by the sequential scheduling strategies in BP decoding of LDPC codes,we propose two improved RCM demodulation algorithms,namely,the GS-BP demodulation algorithm which is based on the standard sequential scheduling strategies and the BS-RBP demodulation algorithm which is based on the informed dynamic scheduling strategies.In the GS-BP demodulation algorithm,message updates are divided into groups according to the inherent order of the bit nodes.In each round of iteration,message updates are performed group by group,which makes the latest updated messages in the previous groups fully utilized in the message update process of the subsequent groups of the current iteration,which helps to speed up the convergence rate of the demodulation algorithm.According to the theory of RBP algorithm,the greater the message residual,indicating the lower the reliability of this message,passing the message with greatest residual preferentially will help to reduce the probability of decoding or demodulation errors.In our proposed BS-RBP demodulation algorithm,residuals of the messages transmitted from bit nodes to symbol nodes(B2S)are regarded as metrics of the reliabilities of messages,and the priorities of the message updates are differentiated according to these metrics.This asynchronous message update strategy facilitates the best use of the latest messages in iterations.Simulation results show that compared with the conventional BP demodulation algorithm of RCM,both of the two proposed demodulation algorithms can significantly accelerate the convergence rate of demodulation.