LAUSR

Sparse code multiple access (SCMA) has been proposed as a candidate air interface (AI) technique for 5G wireless networks. However, the existing resource management schemes with predesigned SCMA codebooks cannot fully exploit user diversities in the frequency domain, thus degrading the performance of SCMA systems. To fully exploit the potential of SCMA, in this paper, we design a more flexible and configurable SCMA through adaptively adjusting the codebook design and assignment according to the user’s features. Specifically, for the uplink networks, first we formulate a detection complexity minimization problem by jointly considering the codebook design (i.e., mapping matrix and constellation graph design) and codebook assignment, which is an integer linear program and NP-hard in general. To tackle this hard problem effectively, first we borrow the idea of dual coordinate search to devise a suboptimal but computational efficient algorithm to determine the mapping matrix and codebook assignment. Based on the obtained mapping matrix, we use the multi-dimensional modulation characteristic of SCMA to carefully design the constellations for each codebook to further reduce the detection complexity. For the downlink networks, we formulate a total power consumption minimization problem by jointly considering the codebook design and assignment and power allocation. Exploiting the special structure of the problem, we employ the Lagrangian dual decomposition technique to propose a fast iterative algorithm, which can solve the problem optimally with low complexity. Finally, we present extensive simulations to exhibit the performance improvement against other algorithms in terms of detection complexity and power consumption. The modified SCMA in this paper can be intelligently optimized based on service and user awareness, which can provide some guidelines for the design of software-defined AI in future wireless networks.