Shannon Theory
S.9.3
Lecture
Information Measures I

On the Trackability of Stochastic Processes Based on Causal Information

Baran Tan Bacinoglu, Yin Sun, Elif Uysal-Biyikoglu

Date & Time

01:00 am – 01:00 am

Abstract

We consider the problem of tracking an unstable stochastic process $X_t$ by using causal knowledge of another stochastic process $Y_t$. We obtain necessary conditions and sufficient conditions for maintaining a finite tracking error. We provide necessary conditions as well as sufficient conditions for the success of this estimation, which is defined as order $m$ moment trackability. By-products of this study are connections between statistics such as R\'{e}nyi entropy, Gallager's reliability function, and the concept of anytime capacity.


Presenters

Baran Tan Bacinoglu

Middle East Technical University

Yin Sun

Auburn University

Elif Uysal-Biyikoglu

Middle East Technical University

Session Chair

Mokshay Madiman

University of Delaware