Access Type

Open Access Dissertation

Date of Award

January 2018

Degree Type

Dissertation

Degree Name

Ph.D.

Department

Computer Science

First Advisor

Hongwei Zhang

Abstract

Predictably reliable communication in wireless networked sensing and control systems (WSC) is a basic enabler for performance guarantee. Yet current research efforts are either focus on maximizing throughput or based on inaccurate interference modelling methods, which yield unsatisfactory results in terms of communication reliability. In this dissertation, we discuss techniques that enable reliable communication in both traditional wireless sensor networks and highly mobile inter-vehicle communication networks. We focus our discussion on traditional wireless sensor networks in Chapter 2 where we discuss mechanisms that enable predictable and reliable communications with no centralized infrastructures. With the promising results in Chapter 2, we extend our methods to inter-vehicle communication networks in Chapter 3. We focus on the broadcast communication paradigm and the unique challenges in applying the PRK interference model into broadcast problems in highly mobile inter-vehicle communication networks. While Chapter 2 and Chapter 3 focus on average reliability, we switch our problem to a more challenging aspect: guaranteeing short-term per-packet reception probability in Chapter 4.

Specifically, we describe the PRKS protocol in Chapter 2 which considers unicast transmission paradigm in traditional static wireless sensor networks. PRKS uses the PRK interference model as a basis for interference relation identification that captures characteristics of wireless communications. For communication reliability control, we design a controller that runs at each link receiver and is able to control the average link reliability to be no lower than an application requirement as well as minimizing reliability variation. We further evaluate PRKS with extensive ns-3 simulations. The CPS protocol described in Chapter 3 considers an one-hop broadcast problem in multi-hop inter-vehicle communication networks. We analyze the challenges of applying the PRK model in this particular setting and propose an approximated PRK model, i.e., gPRK model, that addresses the challenges. We further design principles that CPS uses to instantiate the gPRK model in inter-vehicle communications. We implement the CPS scheduling framework in an integrated platform with SUMO and ns-3 to evaluate our design.

In Chapter 4, we conservatively estimate the background interference plus noise while nodes are receiving packets. In the meantime, receivers decide minimum power levels their sender should use and feedback their decisions to their senders. Senders fuse feedbacks and choose a power level that guarantees expected packet reception probability at each receivers’ side. We notice in our evaluation that guaranteeing short-term reliability causes extra concurrency loss.

Download

Share

COinS