Equivalence for Networks with Adversarial State
Oliver Kosut, Joerg Kliewer

TL;DR
This paper establishes equivalence between noisy networks with adversarial channels and noiseless bit-pipes, revealing how feedback and common randomness can significantly impact network capacity in the presence of adversaries.
Contribution
It provides the first exact equivalence results for networks with compound and arbitrarily varying channels, highlighting the impact of feedback and shared randomness on capacity.
Findings
Equivalence between certain noisy channels with adversaries and noiseless bit-pipes.
Feedback and common randomness can increase the effective capacity.
Small capacity edges can significantly alter network capacity.
Abstract
We address the problem of finding the capacity of noisy networks with either independent point-to-point compound channels (CC) or arbitrarily varying channels (AVC). These channels model the presence of a Byzantine adversary which controls a subset of links or nodes in the network. We derive equivalence results showing that these point-to-point channels with state can be replaced by noiseless bit-pipes without changing the network capacity region. Exact equivalence results are found for the CC model, and for some instances of the AVC, including all nonsymmetrizable AVCs. These results show that a feedback path between the output and input of a CC can increase the equivalent capacity, and that if common randomness can be established between the terminals of an AVC (either by feedback, a forward path, or via a third-party node), then again the equivalent capacity can increase. This leads…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
