基于混合优化策略的下行SCMA码本设计

Downlink SCMA Codebook Design Based on a Hybrid Optimization Strategy

摘要: 稀疏码多址(Sparse Code Multiple Access, SCMA)是一种很有前途的非正交多址接入技术，可以在未来的机器型通信网络中实现大规模连接和高频谱效率。然而，其性能在很大程度上取决于设计良好的多维码本。在本文中，我们提出了一种基于混合优化策略的下行链路码本设计方案，该方案可以在下行瑞利衰落信道上实现近乎最优的性能。通过检查成对错误概率(Pairwise Error Probability, PEP)，我们首先推导了稀疏码本在下行信道中基于Chiani界的符号误码率性能，并将其作为码本优化的设计准则。然后，使用球面打包(Sphere Packing, SP)的方法对母码本进行初始化。接下来，通过混合优化策略，即遗传顺序二次规划法(GA-SQP)得到增强后的母星座(Mother Constellation, MC)。最后通过旋转、交织获得多维码本。仿真结果表明，在下行瑞利衰落信道中，所提出的码本相较于Chen码本和Star-QAM码本具有更优的误码性能。

Abstract: Sparse code multiple access (SCMA) is a promising non-orthogonal multiple access technology that can enable massive connectivity and high spectral efficiency in future machine-type communication networks. However, its performance is heavily dependent on well-designed multi-dimensional codebooks. In this paper, we propose a downlink codebook design scheme based on a hybrid optimization strategy, which achieves near-optimal performance over downlink Rayleigh fading channels. By examining the Pairwise Error Probability (PEP), we first derive the Symbol Error Rate (SER) performance of the sparse codebook over downlink channels based on the Chiani bound, and establish it as the design criterion for codebook optimization. Then, the mother codebook is initialized using the sphere packing method. Subsequently, an enhanced mother constellation is obtained through a hybrid optimization strategy, namely a Genetic Sequential Quadratic Programming method. Finally, the multi-dimensional codebook is generated through rotation and interleaving. Simulation results demonstrate that the proposed codebook achieves performance gains over state-of-the-art codebooks in downlink Rayleigh fading channels.