Compressed sensing based multiuser detection for sparse code multiple access

Since the current orthogonal multiple access techniques do not fulfill the requirements of the new generation communication technologies, non-orthogonal multiple access techniques have been developed. Sparse code multiple access, which is one of the non-orthogonal multiple access techniques based on the idea of transmit with more users over the same time-frequency resources, is estimated to be used especially in machine type communication with sparse codebooks and receiver with near-optimal performance. But with massive machine-type communications and the internet of things, the number of users will increase enormously in the near future. Message passing algorithm, which is the receiver of the sparse code multiple access technique; it has a structure whose complexity increases exponentially as the number of users increases. In addition, only a small part of the users actively use the system in machine-type communication. Therefore, it can be said that there is a sparsity, as only a small part of all users are active in a time slot. Active user detection and analyzing the data of these users reduces the computational load. Based on these reasons, two methods have been developed to reduce the complexity of the receiver structure and to detect active users and estimate the channel. In the first method developed, a method in addition to the receiver structure MPA was proposed by using the compressed sensing theory. Sparse errors that occur after initial detection, where several iterations occur, are reconstructed using compressed sensing theory. The performance of the receiver is increased with less number of iterations with this method. In the second method developed, an active user detection and channel estimation method that is dynamic and adaptive to sparsity has been developed by using the compressed sensing theory. The developed method has the best performance in cases where the sparsity is high compared to the important methods based on compressed sensing. The method is applicable as it does not need prior sparsity knowledge.