Decentralized multi-user multi-armed bandits with user dependent reward distributions

The uncoordinated spectrum access problem is studied using a multi-player multi-armed bandits framework. We consider a decentralized multi-player stochastic multi-armed bandit model where the players cannot communicate with each other and can observe only their own actions and rewards. Furthermore, the environment may appear differently to different players, i.e., the reward distributions for a given arm may vary across players. Knowledge of time horizon T is not assumed. Under these conditions, we consider two settings - zero and non-zero reward on collision (when more than one player plays the same arm). Under the zero reward on collision setting, we present a policy that achieves expected regret of O(log T) over a time horizon of duration T. While settings with non-zero rewards on collisions and varying reward distributions of arms across players have been considered separately in prior work, a model allowing for both has not been studied previously to the best of our knowledge. With this setup, we present a policy that achieves expected regret of order O(log^{2 + \delta} T) for some 0 < \delta < 1 over a time horizon of duration T.

Name thesis:degree_name

M.S.

Level thesis:degree_level

Thesis

Discipline thesis:degree_discipline

Electrical & Computer Engr

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2020

Author dc:creator

• Magesh, Akshayaa

Contributors dc:contributor

• Veeravalli, Venugopal V.

• multiarmed bandits, multi-player, spectrum access, decentralized

Statement dc:rights

• Copyright 2020 Akshayaa Magesh

Language dc:language

en

Handle dc:identifier

http://hdl.handle.net/2142/107917

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/107917