SPARCs for Unsourced Random Access

Unsourced random-access (U-RA) is a type of grant-free random access with virtually unlimited number users, which only certain K a are active on the same time slot. Users employ exactly codebook, and task receiver to decode list transmitted messages. We present concatenated coding construction for U-RA AWGN channel, in sparse regression code (SPARC) used as an inner create effective outer OR-channel. Then resolve multiple-access interference OR-MAC. propose modified version approximate message passing (AMP) algorithm decoder give precise asymptotic analysis error probabilities AMP hypothetical optimal MAP decoder. This shows that under decoding can achieve vanishing per-user probability limit large blocklength users at sum-rates up symmetric Shannon capacity, i.e. long R <; 0.5log xmlns:xlink="http://www.w3.org/1999/xlink">2 (1+K SNR). extends previous point-to-point optimality results about SPARCs unsourced multiuser scenario. Furthermore, we optimization find power allocation SPARC minimizes SNR required given target