1 / 8

A Continuity Theory of Source Coding over Networks WeiHsin Gu, Michelle Effros, Mayank Bakshi, and Tracey Ho FLoWS PI

A Continuity Theory of Source Coding over Networks WeiHsin Gu, Michelle Effros, Mayank Bakshi, and Tracey Ho FLoWS PI Meeting, Washington DC, September 2008. Status Quo. Achievable rate regions are hard to characterize. Only solved for some small networks and multicast networks

didrika
Download Presentation

A Continuity Theory of Source Coding over Networks WeiHsin Gu, Michelle Effros, Mayank Bakshi, and Tracey Ho FLoWS PI

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. A Continuity Theory of Source Coding over Networks WeiHsin Gu, Michelle Effros, Mayank Bakshi, and Tracey Ho FLoWS PI Meeting, Washington DC, September 2008

  2. Status Quo Achievable rate regions are hard to characterize. • Only solved for some small networks and multicast networks • Inner (achievable) and upper (converse) bounds are derived for some example networks Open questions: • Are those earlier derived bounds tight? • Does single-letter characterization always exist?

  3. New Insights • Develop and investigate some abstract properties of the achievable rate regions • As functions of probability distribution and distortion vector, are the achievable rate regions continuous? Motivations • Understand the existence of 1-letter characterizations • Applications: estimation of achievable rate regions Why is it hard? • Block length is unbounded in the definition of the achievable rate regions • Don’t know much about the achievable rate regions

  4. Formulation and Notation • Define a family of network source coding problems which contains the prior example networks as special cases. The family contains four subfamilies • Lossless non-functional • Lossy non-functional • Lossless functional • Lossy functional • : Source : Side information • : Probability distribution : Distortion vector • : Lossy rate region • : Lossless rate region

  5. Main Results • is concave in • is continuous in when • For any non-functional source coding problem, is continuous in for all • For super-source networks, both and are continuous in for all

  6. Continuity w.r.t. • is continuous in if and only if • is upper semi-continuous in • is lower semi-continuous in • We show that the achievable rate regions are inner semi-continuous in for • Lossless non-functional • Lossy non-functional • Lossy functional

  7. A Possible Solution • We conjecture that outer semi-continuity w.r.t. is true when random variables have finite alphabets • A possible approach: for any network , consider the super-source network • If the above equality is true, then is upper semi-continuous in

  8. Next-Phase Goals • Prove or disprove upper semi-continuity w.r.t. for general networks. • Characterize a larger family of networks where upper semi-continuity holds • Investigate some other useful abstract properties • Understand the existence of achievable rate regions

More Related