Some Results on the Computation of Feedback Capacity of Gaussian Channels with Memory

We study the problem of computing the capacity of channels with feedback for the class of Gaussian channels with linear state-space models (possibly with hidden states) under the presence of quadratic input constraints. We first show that the input-output directed information is maximized by a Gaussian feedback policy. A few special cases are considered where such an optimization can be performed in a computationally tractable manner. In such cases, we show that the supremum of directed information admits a single-letter expression involving the convex log-determinant program as the block length tends to infinity.

• Publication year

  2018

• Venue

  Allerton Conference on Communication, Control, and Computing

• Publication date

  2018-10-01

• Fields of study

  Mathematics, Computer Science

• Identifiers
  DOI 10.1109/ALLERTON.2018.8636014
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• Hierarchical optimality of linear controllers-encoders-decoders operating at control-coding capacity of LQG control systems

  2017cited by this paper
• Sequential Necessary and Sufficient Conditions for Capacity Achieving Distributions of Channels With Memory and Feedback

  2016cited by this paper
• Semidefinite representation of sequential rate-distortion function for stationary Gauss-Markov processes

  2015cited by this paper
• LQG Control With Minimum Directed Information: Semidefinite Programming Approach

  2015cited by this paper
• Information Structures for Feedback Capacity of Channels with Memory and Transmission Cost: Stochastic Optimal Control & Variational Equalities-Part I

  2015cited by this paper
• The Capacity of Channels With Feedback

  2006influential reference
• Feedback Capacity of Stationary Gaussian Channels

  2006cited by this paper
• Finite State Channels With Time-Invariant Deterministic Feedback

  2006cited by this paper
• Elements of Information Theory

  2005cited by this paper
• Directed information for channels with feedback

  1998cited by this paper
• Determinant Maximization with Linear Matrix Inequality Constraints

  1998cited by this paper
• CAUSALITY, FEEDBACK AND DIRECTED INFORMATION

  1990cited by this paper
• Gaussian feedback capacity

  1989cited by this paper
• A coding scheme for additive noise channels with feedback-I: No bandwidth constraint

  1966cited by this paper
• Sequential transmission using noiseless feedback

  1963cited by this paper
• Ieee Transactions on Information Theory 2 on the Feedback Capacity of Power Constrained Gaussian Noise Channels with Memory

  year unknowncited by this paper

• Communication over LQG Control Systems: A Convex Optimization Approach to Capacity

  2025influential citation
• Neural Estimation and Optimization of Directed Information Over Continuous Spaces

  2022cites this paper
• Feedback Capacity of Parallel ACGN Channels and Kalman Filter: Power Allocation with Feedback

  2021cites this paper
• A Connection between Feedback Capacity and Kalman Filter for Colored Gaussian Noises

  2020cites this paper
• Feedback Capacity and a Variant of the Kalman Filter with ARMA Gaussian Noises: Explicit Bounds and Feedback Coding Design

  2020cites this paper
• Bidirectional Information Flow and the Roles of Privacy Masks in Cloud-Based Control

  2019cites this paper