Work

Coding for Downlink Massive Random Access

Year: 2024
Type: dissertation
Abstract: Consider a massive access scenario consisting of a central base-station (BS) and n potential users.The BS aims to communicate sources (X1, · · · , Xk) to a randomly activated subset of k ≪ n users via a downlink common message. If each source were to be transmitted alongside the identity of it’s intended user, H(X1, · · · , Xk) + k log(n) bits would be required, since the cost of specifying a user identity is log(n) bits. This thesis shows that if the distribution of the sources is exchangeable, then the k log(n) bit overhead can be reduced to O(k) bits, with improvement to O(log(k)) bits in certain cases. A connection is then drawn between downlink massive random access and the study of finite exchangeable sequences. Through this connection, a new scaling optimal relative entropy version of the finite de Finetti theorem is developed. Lastly, a lossy coding version of this problem is studied.
Author: Song, Ryan
Language: English
Open Access status: green
OpenAlex ID: W7133102177
Cites: 24
Cited by:
Related to: 10