Private Broadcasting over Independent Parallel Channels

TL;DR

This paper characterizes the capacity region for private broadcasting over independent parallel channels, demonstrating the optimality of superposition strategies and Gaussian inputs, with extensions to fading channels showing significant gains.

Contribution

It establishes the capacity region for private broadcasting over parallel channels and proves the optimality of superposition coding and Gaussian inputs, extending to fading channels.

Findings

Capacity region characterized for parallel channels.

Superposition coding is optimal for this setting.

Gaussian inputs are optimal for Gaussian channels.

Abstract

We study private broadcasting of two messages to two groups of receivers over independent parallel channels. One group consists of an arbitrary number of receivers interested in a common message, whereas the other group has only one receiver. Each message must be kept confidential from the receiver(s) in the other group. Each of the sub-channels is degraded, but the order of receivers on each channel can be different. While corner points of the capacity region were characterized in earlier works, we establish the capacity region and show the optimality of a superposition strategy. For the case of parallel Gaussian channels, we show that a Gaussian input distribution is optimal. We also discuss an extension of our setup to broadcasting over a block-fading channel and demonstrate significant performance gains using the proposed scheme over a baseline time-sharing scheme.