Ph.D. Research Proposal: Nan Xie
Wednesday, November 20, 2013
11:00 AM-1:00 PM
Title: Towards a Better Cellular Network: Stability, Delay Performance and Scheduling
Advisor: Dr. Steven Weber
Date: Wednesday, November 20, 2013
Time: 11:00 a.m.
Location: ECE Conference Room 303, 3rd Floor, Bossone Research Enterprise Center
This thesis proposal addresses several issues of design and performance analysis in cellular communication networks. We investigate topics relevant to both uplink and downlink. For the uplink part, we address the stability region of the slotted Aloha medium access control protocol under the collision channel model, for the case of a finite number of independent users. The stability region (i.e., the set of arrival rate vectors such that the whole queueing system can be made stable) is in general unknown when the number of users is more than three. We seek to characterize the set of stabilizable rate vectors, whereas most existing work only provides bounds of the region of stabilized rate vectors for given contention probability vectors.
For the downlink part, we seek to characterize the delay when broadcasting random linear combinations of the information packets over independent erasure channels to a finite number of users. Of interest is the random delay until all the receivers recover all the packets initially queued at the base station (i.e., the sender). This falls into the study of certain order statistic of random variables. We obtain tight lower and upper bounds, exact expressions and finite-step computational procedures (recurrence) for the moment(s) of the random delay. We also investigate the dependence of the delay on the code blocklength (under random linear combinations of packets as the scheme employed in random linear network coding), and on the number of receivers, respectively. Our proposed work includes the following two topics. First, we propose to study the tradeoff between increasing the blocklength (to reduce the overall delay) and meeting the delivery deadline requirements of individual packets. Second, we will also be interested in the scenario where receivers already have a subset of the packets (i.e., as side information) hosted at the base station, and are interested in only the remaining packets. Most of the work is related to network coding, which for more than one and a half decade, has successfully generated theoretical and practical interests in information theory, communication theory and networking communities.