English

Exact minimum number of bits to stabilize a linear system

Systems and Control 2021-11-25 v3

Abstract

We consider an unstable scalar linear stochastic system, Xn+1=aXn+ZnUnX_{n+1}=a X_n + Z_n - U_n, where a1a \geq 1 is the system gain, ZnZ_n's are independent random variables with bounded α\alpha-th moments, and UnU_n's are the control actions that are chosen by a controller who receives a single element of a finite set {1,,M}\{1, \ldots, M\} as its only information about system state XiX_i. We show new proofs that M>aM > a is necessary and sufficient for β\beta-moment stability, for any β<α\beta < \alpha. Our achievable scheme is a uniform quantizer of the zoom-in / zoom-out type that codes over multiple time instants for data rate efficiency; the controller uses its memory of the past to correctly interpret the received bits. We analyze its performance using probabilistic arguments. We show a simple proof of a matching converse using information-theoretic techniques. Our results generalize to vector systems, to systems with dependent Gaussian noise, and to the scenario in which a small fraction of transmitted messages is lost.

Keywords

Cite

@article{arxiv.1807.07686,
  title  = {Exact minimum number of bits to stabilize a linear system},
  author = {Victoria Kostina and Yuval Peres and Gireeja Ranade and Mark Sellke},
  journal= {arXiv preprint arXiv:1807.07686},
  year   = {2021}
}

Comments

Extended version of the paper accepted to IEEE Transactions on Automatic Control

R2 v1 2026-06-23T03:08:09.486Z