Research And Applications Of Spinal Codes Based On Massive Machine Type Communication

Massive machine type communication is a very important application scenario of the 5G standard.It has the characteristics of small transmission data,high access delay,large number of devices,and long device life cycle.Spinal codes are a kind of classical rateless codes.By setting the initial state and information segmentation,the hash function is used to serially generate the theoretically approximate infinite coding states,then the information is transmitted through passes.When receiving transmitted information,the decoding process continuously codes the candidate sequence and sends the decoding feedback information in time until the decoding is successful.The rateless spinal codes have a simple coding structure and excellent rate performance.It is theoretically proved that the channel capacity is reachable on the BSC and AWGN channels.More importantly,spinal codes can adapt to various channel states(even if the channel condition is unknown).Meanwhile,even the messages are short,it could approach the capacity.Due to the natural conditions of the coding structure,spinal codes still have many problems.Generally speaking,because of the unequal error protection property resulting from the serial structure,the error correction performance of the spinal codes is usually poor.The decoding complexity of spinal codes is relative high in rateless transmission.Up to now,there is not a perfect parallel decoding scheme for spinal codes.We focus on the improvement of the error performance on spinal codes,then study the application in the massive machine type communication.The main work includes the following contents: A partial cascaded spinal codes are designed to improve the problem of poor error correction performance in original spinal codes.By analyzing the error positions of the original spinal codes,we use the outer codes,which show excellent error correction performance,to provide more protection for the information bits with higher error probability.Then we cascade the information bits of the original spinal codes and the output of the outer codes.The loop check matrix is designed to correct errors.According to the unequal error protection property,the unequal length code symbol generated method is used to improve the number of highly reliable coded information bits.We also study the characteristics the burst erroring order to design the outer code parameters,which must adapt to the coding structure and transmission scheme of the spinal codes.Aiming at the problem of weaky error correction performance of original spinal codes,an iterative list decoding algorithm is designed.Spinal codes use iterative decoding of list decoding algorithm between inter-pass and pass-to-pass.Based on this algorithm and partial cascaded spinal codes,a joined-list partial cascaded decoding algorithm of spinal codes is proposed.Firstly,the algorithm decodes the transmission information through the inter-pass list decoding.Then it combines the outer code's syndrome check messages and the maximum zero distance to correct the burst error.Next,algorithm iteratively executes the pass-to-pass list decoding algorithm.Finally,it verifies the output.We study finite length performance of the partial cascading spinal codes in the AWGN channel,and derive its theoretical analysis on upper bound of error probability.The simulation results show that joined-list partial cascaded decoding algorithm of spinal codes may be widely used in 5G applications——the mMTC scenarios,for example,smart meter,smart home and industrial automatio.