Clement Nthambazale Nyirenda

MSc(Eng) Candidate

Address:
School of Electrical, Electronic and Computer Engineering
University of KwaZulu-Natal
King George V Avenue, Durban, South Africa, 4041

Room:4-10
Office: +27 (031) 260 2736

Cell: +27 072 140 4564
Fax: +27 (031) 260 2740
E-mail: Nyirendac{at}ukzn.ac.za


    Short CV       Research       Other Interests        Publications

Short CV

I am a second year MSc (Eng) student at the Radio Access Technologies (RAT) Centre in the School of Electrical, Electronics and Computer Engineering, University of KwaZulu-Natal, Durban, South Africa. I received my Diploma and BSc. degree in Electrical Engineering, in 1998 and 2000, respectively, both from the University of Malawi. From September 2000 to August 2004, I worked as a Staff associate/Assistant Lecturer in Electronics and Computer Engineering in the Department of Electrical Engineering at the Polytechnic, a constituent college of the University of Malawi.

In August 2004, I received a Fellowship from the African Network of Scientific and Technological Institution (ANSTI).This Fellowship enabled me to join the University of KwaZulu-Natal for a two-year MSc(Eng) degree in Computer Engineering by research. I am now in the final stages of my research.

Research

"So far as the laws of mathematics refer to reality, they are not certain. And as far as they are certain they do not refer to reality." Albert Einstein

Speaking in broad terms, my research life is motivated by the passion to develop systems that are self-cofiguring, self-organizing, self-learning, self-adaptive etc. The need for such systems becomes greater and greater as the world is becoming more and more complex. Research has shown that traditional analytical mathematical techniques are failing to cope with these complexities. It is quite encouraging to observe that the science world is now trying to mimick nature (in other words God) in order to solve the problems posed by the complex situations of this world. A recently emerging field in this aspect is widely known as Computational Intelligence (CI). The main components of CI are neural networks, fuzzy set technology and evolutionary computation. Each one of them plays an important role in this triumvirate.

At present, I am specifically involved in the application of CI in Communication systems. One issue that is proving to be a hard nut to crack is that of Internet Congestion Control. Internet congestion control is composed of three processes: 1) Congestion detection at a gateway e.g. router; 2) Generation and transmission of congestion notification signal to the traffic sources; 3) End-host algorithms (e.g. TCP) which control the flow of traffic. Although my work touches all the three areas, it has largely addressed the first two.

Recently, fuzzy logic has been used for the detection of congestion in the router. Although the performance of fuzzy logic based congestion detection algorithms is generally better than that of traditional algorithms i.e. RED, PI, PID, there still exists ample room for improvement. In my research I have studied the deficiencies of the existing fuzzy algorithms and proposed a Fuzzy Logic Congestion Detection (FLCD) which synergistically combines the good characteristics of the fuzzy approaches with those of the traditional approaches. The membership functions (MFs) of the FLCD algorithm are designed automatically by using Multi-objective Particle Swarm Optimization (MOPSO), a population based stochastic optimization algorithm which falls under evolutionary computation in CI’s triumvirate. This optimization process enables the FLCD to achieve optimal performance on all the major objectives of Internet congestion control. I have also designed and implemented a self-learning and adaptation mechanism for the FLCD algorithm.

On the generation and transmission of congestion notification signal, I have designed and implemented a fuzzy logic based dual explicit congestion notification algorithm. This algorithm combines the merits of the Explicit Congestion Notification (ECN) and the Backward Explicit Congestion Notification (BECN) mechanisms. Under ECN, a router marks packets in their forward path from sender to receiver. Upon receipt of a congestion marked packet, the TCP receiver informs the sender (in the subsequent ACK) about incipient network congestion. In response, the sender invokes the congestion avoidance algorithm while under BECN, a router uses the Internet Control Message Protocol (ICMP) Source Quenches as a means for reverse congestion notification. ECN is more reliable than BECN because it uses ACKs. On the other hand, BECN is much faster than ECN because the congestion signal does not traverse the round trip distance before the TCP sender reacts to it. As a result, BECN tremendously reduces packet transfer delays, delay variations and packet losses due to buffer overflows. BECN’s drawbacks relate to the extra overhead required for the generation of ICMP Source Quenches (ISQ). I have proposed a mechanism which reduces the generation of ISQs significantly while maintaining all the good attributes of BECN. All this work is done on the Network Simulator (version 2.28) platform. The outputs of this work can be viewed under the Publications section.

My future research interests include Type-2 Fuzzy Logic, Swarm Intelligence, Evolutionary Multi-objective optimization, Wireless Sensor networks, Evolvable Hardware, Reconfigurable Hardware – FPGAs.

Other Interests

Work related

  • GNU/Linux Programming
  • Computer Networking
  • Voice over IP

Life in general

Publications

 

Last updated on 4 August 2006.

 

You are visitor number

Back to UKZN School of Electrical, Electronic and Computer Engineering Page