NeTS: Small: Principles and Protocols for Traffic-Insensitive Performance in Wireless Networks

点击翻译按钮获取中文摘要

The rapid growth of smart phones as well as other intelligent wireless devices generates significant amount of diversified traffic ranging from sending warning messages to watching online videos over wireless networks such as cellular networks, Internet of Things, and cyber-physical systems. Such wide range of traffic calls for the development of wireless algorithms with traffic-insensitive performance in the sense that the network performance (such as throughput and latency) does not depend on the distribution of network traffic except the mean traffic load. This is tremendously valuable in the presence of highly heterogeneous wireless applications. The goal of this project is to enable the wireless algorithm design to be robust to the diversified service demands and meet the stringent performance needs of future wireless systems. This research project will also be closely integrated with PI's undergraduate and graduate courses, and will provide hand-on experiences for undergraduate and high school students in wireless networking technology.This project aims to establish efficient, adaptable and scalable algorithms that are robust to the diverse wireless traffic penetrating almost every wireless system. This calls for new dynamic algorithms and analytical framework for both delay and short-term fairness insensitivity, which are extremely important for real-time applications. Towards this end, the project will answer the following eight questions that address key elements and challenges of popular wireless systems: (1) For a realistic model of flow arrivals and departures, is there an algorithm that achieves good throughput, delay and short-term fairness insensitivity? (2) How can such an algorithm be implemented in a multi-channel base station? (3) How can such an algorithm be implemented in the uplink with limited coordination by the base station? (4) Is it possible to implement such an algorithm in a purely distributed way? (5) What are good mathematical models to study the delay insensitivity of such algorithms to flow-size distributions? (6) Does the algorithm have good short-term fairness performance? (7) What are good models for arrivals and departures of flows and how do they affect the performance of the algorithm? (8) How does one validate the algorithms using laboratory experiments? The novelty in the proposed research lies in the development of efficient, robust and easily implementable scheduling algorithms in various wireless systems as well as the methodology for analyzing both delay and short-term fairness insensitivity.