Understanding and Resolving Wireless Collision with PHY Techniques

The Hong Kong University of Science and Technology
Department of Computer Science and Engineering


PhD Thesis Defence


Title: "Understanding and Resolving Wireless Collision with PHY 
Techniques"

By

Mr. Xiaoyu JI


Abstract

Due to the broadcast nature and lack of collision detection mechanism, 
wireless networks suffer from collision. Collision happens when two or 
more data packets overlap in the time domain at receiver and none of them 
can be received correctly. Collision increases packet delivery delay, 
decreases network throughput and incurs extra energy cost because of 
retransmissions. To tackle collision, researchers propose abundant 
protocols in Medium Access Control (MAC) layer. The essential idea behind 
these protocols is to properly coordinate multiple senders to access a 
shared channel and avoid the case where there are two or more senders 
accessing a shared channel simultaneously. While in recent years, an 
increasing number of protocols are invented to approach collision in the 
physical layer (PHY). PHY layer demonstrates promising properties, e.g., 
power indicator like RSSI, the effect of capture and constructive 
interference, which enable us resolve collision from its nature as 
wireless signals.

This thesis first draws a new logical roadmap along which we understand 
and deal with collision. The roadmap is: collision avoidance -> collision 
tolerance -> collision cancellation -> collision exploitation. We in the 
beginning avoid collision out of its harm, and then we can just tolerate 
them with some "powerful weapons". After that, we try to cancel collision 
with more powerful techniques and finally we realize that collision could 
be utilized to assists us. Guided by this roadmap, the thesis mainly 
addresses collision to achieve energy-efficient scheduling and 
near-optimal channel utilization, respectively in the stages of collision 
avoidance and tolerance. In the stages of cancellation and exploitation, 
we recover collided signals by signal cancellation and make use of 
collision patterns and even generate collision to build a decoupled 
control plane based on the data plane.


Date:			Wednesday, 24 June 2015

Time:			10:30am - 12:30pm

Venue:			Room 2132C
 			Lift 19

Chairman:		Prof. Jang Kyo Kim (MAE)

Committee Members:	Prof. Ke Yi (Supervisor)
 			Prof. Gary Chan
 			Prof. Lei Chen
 			Prof. Jiang Xu (ECE)
 			Prof. Jianping Wang (CityU)


**** ALL are Welcome ****