Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
149 tokens/sec
GPT-4o
7 tokens/sec
Gemini 2.5 Pro Pro
45 tokens/sec
o3 Pro
4 tokens/sec
GPT-4.1 Pro
38 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

Optimal Feedback Communication via Posterior Matching (0909.4828v2)

Published 28 Sep 2009 in cs.IT and math.IT

Abstract: In this paper we introduce a fundamental principle for optimal communication over general memoryless channels in the presence of noiseless feedback, termed posterior matching. Using this principle, we devise a (simple, sequential) generic feedback transmission scheme suitable for a large class of memoryless channels and input distributions, achieving any rate below the corresponding mutual information. This provides a unified framework for optimal feedback communication in which the Horstein scheme (BSC) and the Schalkwijk-Kailath scheme (AWGN channel) are special cases. Thus, as a corollary, we prove that the Horstein scheme indeed attains the BSC capacity, settling a longstanding conjecture. We further provide closed form expressions for the error probability of the scheme over a range of rates, and derive the achievable rates in a mismatch setting where the scheme is designed according to the wrong channel model. Several illustrative examples of the posterior matching scheme for specific channels are given, and the corresponding error probability expressions are evaluated. The proof techniques employed utilize novel relations between information rates and contraction properties of iterated function systems.

Citations (163)

Summary

  • The paper introduces posterior matching, a novel principle for feedback communication over memoryless channels that achieves rates near mutual information while simplifying transmission.
  • This framework provides a unified understanding for existing feedback schemes like Horstein for BSC and Schalkwijk-Kailath for AWGN channels, demonstrating its broad applicability.
  • Under regularity conditions, the posterior matching scheme is proven capacity-achieving, with error probabilities tied to the convergence rates of associated iterated function systems.

Optimal Feedback Communication via Posterior Matching

The paper "Optimal Feedback Communication via Posterior Matching" by Ofer Shayevitz and Meir Feder explores a new principle for communication over memoryless channels with noiseless feedback, termed "posterior matching." The authors propose a transmission scheme based on this principle that seamlessly integrates feedback into the communication process, simplifying the transmission strategy while achieving rates close to mutual information. This paradigm provides a unified approach that encompasses well-known schemes, such as the Horstein scheme for Binary Symmetric Channels (BSC) and the Schalkwijk-Kailath scheme for Additive White Gaussian Noise (AWGN) channels.

Overview of the Posterior Matching Principle

Feedback in memoryless channels traditionally does not increase capacity, yet it can greatly simplify transmission schemes to achieve capacity. Posterior matching utilizes the feedback to align the posterior distribution of the message with the input distribution required by the channel. In essence, the transmitted signal should represent the information missing at the receiver in an optimal way based on the observed outputs. This framework allows for a recursive representation of transmission functions, making the system run effectively "on the fly" with minimal computational complexity.

Recursive Scheme Representation

The authors present a recursive characterization of the posterior matching scheme using the inverse c.d.f. of the input and the inverse channel's distribution function. The recursive formulation is particularly useful for continuous distributions, where it represents a simple yet effective modeling of communication. It shows that the feedback mechanism can emulate the desired input/output distributions, offering a simplified yet robust transmission strategy that is capacity achieving under general conditions.

Application to Known Feedback Schemes

The paper demonstrates that classical feedback schemes fit within the posterior matching framework. Specifically, it shows that the Horstein scheme for BSC can be interpreted as slicing the posterior distribution at its median, reflecting a binary representation that achieves the required input distribution. Similarly, the Schalkwijk-Kailath scheme for AWGN channels is rephrased as transmitting the MMSE error term with appropriate scaling to maintain the power constraint, highlighting the innate simplicity and effectiveness of such an approach when applied universally.

Achievability and Error Analysis

A core result presented is that under certain regularity conditions, the posterior matching scheme achieves any rate below the mutual information for a wide class of input/channel pairs, whether discrete or continuous. The essence lies in proving that the posterior pdf around the message concentrates sufficiently as transmission progresses, allowing for accurate decoding even at high rates.

Additionally, the authors delve into error probability analysis, demonstrating that convergence rates of iterated function systems (IFS) dictate achievable rates and error probabilities. Such an approach provides explicit insights into the trade-offs involved in feedback communication, offering guidelines on designing systems with lower error performance.

Extensions and Implications

The paper discusses the flexibility of the posterior matching scheme through variants and extensions, such as the inclusion of a different ordering of inputs to handle fixed-point phenomena in certain channels. Furthermore, considerations on channel model mismatch indicate that the scheme maintains some robustness against unanticipated channel changes.

Conclusion and Future Work

This work provides a robust yet straightforward framework for feedback-based communication that neatly ties together various classical schemes while inviting potential future research avenues. Specifically, its application to channels with memory and possible universal adaptations across different channel models are promising areas for further exploration. The innovative use of posterior matching could lead to more effective communication systems exploiting feedback capabilities, enhancing reliability for numerous practical scenarios.