This talk considers communication over the Gaussian multiple-access channel in the regime where the number of users scales linearly with the codelength. I will begin by discussing near-optimal coding schemes for small user payloads, and explain why these cannot be adapted efficiently to larger payloads. I will then describe a coded CDMA scheme for larger payloads, where each user's information is encoded via a linear code before being transmitted using a signature sequence. With an efficient Approximate Message Passing (AMP) decoder that can be tailored to the structure of the linear code, I show that coded CDMA schemes can achieve state-of-the-art performance for payloads up to hundreds of bits, with exact asymptotic performance guarantees. I will also explain how the proposed schemes can be adapted to incorporate random user activity.
This is joint work with Pablo Pascual Cobo, Kuan Hsieh and Ramji Venkataramanan.