PhD Defense of Computer Science Scholar, Mr. Anwar Ahmed Khan

Title: Preemptive Priority Based Data Fragmentation Scheme for Heterogeneous Traffic in Wireless Sensor Networks (WSNs)
Date: Tuesday, Aug 25, 2020
Time: 2:30 p.m.
Venue: VC Room, 1st floor, Faysal Academic Center, IBA City Campus, Karachi, Pakistan.
Advisor: Dr. Sayeed Ghani

National Examiners:
Dr. Muhammad Hamad Alizai (LUMS)
Dr. Bilal Muhammad Khan (NUST)

Abstract
Wireless Sensor Networks (WSN) comprise of tiny devices capable of sensing, computing and communicating the parameters of interest. With the emergence of Internet of Things (IoT) technology, the application areas of WSN have been rapidly broadening. Advanced applications of Wireless Sensor Networks (WSNs) often deal with heterogeneous data; some of such applications include healthcare, industrial automation and control, vehicular ad hoc networks and smart homes. In most application scenarios, different traffic types have specific service requirements in terms of delay tolerance, throughput, bandwidth efficiency and reliability. This implies that WSN nodes should be able to categorize each type of data based on the required priority level. Therefore, while developing the communication protocols for WSN nodes, differentiated service is often offered to facilitate different types of traffic.

Media Access Control (MAC) layer has been regarded as the best choice for designing priority mechanisms for WSN nodes. Various adaptive MAC approaches for WSN have been developed in the past, focusing on prioritizing heterogeneous traffic, such as "adaptive contention", "duty-cycle adaptation" and "queue management". Furthermore, various hybrid schemes have also been proposed which combine several priority mechanisms to have an aggregated effect. Although the existing MAC schemes have been able to reduce the delay for high priority/urgent traffic by suppressing/delaying the transmission of low priority data, none of these schemes have achieved true preemption.

This dissertation proposes a novel data fragmentation scheme for low priority traffic in order to improve the delay performance of high priority traffic in WSN. It is proposed that instead of transmitting the low priority packet as a single unit, it should be transmitted in small fragments with pauses in between. These pauses are termed as "interruptible periods" and have been introduced to allow the high priority traffic to begin transmission, even before the complete transmission of a low priority packet. This scheme has been shown to reduce the delay of high priority data as instead of waiting for the complete low priority packet to end its transmission, the urgent data may be transmitted only after transmission of a fragment. Based on the proposed fragmentation scheme, a novel MAC protocol FROG-MAC has been developed for prioritizing heterogeneous traffic in WSN. The focus of FROG-MAC is to offer pseudo-preemptive priority, which may be considered very close to preemptive priority, to the high priority traffic not only by holding the low priority packets in the queue, but also by interrupting the ongoing transmission of a low priority packet.

The feature of non-overlapping contention window scheme has also been integrated in FROG-MAC. As a result, FROG-MAC promises to bring performance improvements mainly in terms of delay for WSN as compared to the conventional prioritization mechanisms such as IEEE 802.11e. Unlike these schemes, FROG-MAC does not need to achieve synchronization, nor it needs to transmit the scheduling information which results in reduced transmission and processing cost.

FROG-MAC has been fully designed in this dissertation and the operation of the protocol has been described including the details of hidden terminals and collision avoidnce. Two types of traffic: Normal (low priority) and Urgent (High Priority) have been used to demonstrate the design and operation of FROG-MAC. The protocol has been implemented over Contiki Operating System for TelosB/Tmote Sky Platform. Simulations have been performed to conduct performance evaluation of FROG-MAC using maximum delay, average delay and throughput or Packet Delivery Ratio (PDR) as performance parameters. Single hop star topology has been used for conducting simulations as this setting resembles the real-world application scenarios of WSNs generating heterogeneous traffic (such as Body Area and Vehicular Networks). Network size and fragment size have been varied to study the influence over the performance of normal and urgent traffic.

It has been found that when the network size increases, the delay for both normal and urgent traffic increases due to the increased network traffic load, collision and retransmissions. Also, the PDR for both traffic types reduces with increasing network size. Furthermore, for increase in the fragment size, the delay of urgent traffic increases, whereas that of normal traffic reduces. Due to the fragmentation scheme, the delay and PDR of urgent traffic depicted better performance as compared to normal for all experiments. Hence, the simulation results confirmed the trend initially observed using high level MATLAB implementation. In order to validate the performance of FROG-MAC, simulations have been conducted to compare the throughput and delay of urgent traffic with two recent protocols that have implemented similar priority schemes: eMC-MAC and urg-MAC. The results of FROG-MAC depicted performance improvement for both the normal and urgent traffic.