Using Limited Feedback for Communicating in the Presence of Unknown Interference
Krish Eswaran, Anand D. Sarwate, Anant Sahai and Michael Gastpar
Feedback plays a significant role in the design of most practical communication systems, and has also inspired many results in the information theory literature. Information-theoretic results on feedback can be grouped into two categories: the effect of feedback on the capacity or reliability of communication, and the use of feedback to overcome uncertainty about the channel model. In most practical communication systems, physical constraints limit the amount of feedback one can send, and therefore the amount of feedback required is a key concern. Thus, an understanding of how limited feedback affects the above settings provides insights into the design of practical systems.
We show how we to achieve the empirical capacity using a model similar to Shayevitz and Feder's but using a limited feedback strategy. To do this, we adapt the feedback-reducing block/chunk strategies used earlier in the context of reliability. They are in turn inspired by Hybrid ARQ. The flavor of our algorithm is different--in our scheme the decoder uses the feedback link to terminate rounds that are too noisy but otherwise attempts to correct the error in less noisy rounds. By doing away with the output feedback, we lose some of the simplicity of the scheme in Shayevitz and Feder, but we show that a similar performance can still be obtained with almost negligible feedback.
- K. Eswaran, A. D. Sarwate, A. Sahai, and M. Gastpar, "Using Zero-Rate Feedback on Binary Additive Channels with Individual Noise Sequences," Proceedings of the 2007 International Symposium on Information Theory (ISIT 2007), Nice, France, June 2007.