Speaker Details

Speaker 1

Dr. Asad Arfeen

Dr. Asad Arfeen is an Associate Professor at the Department of Computer and Information Systems Engineering of NED University of Engineering & Technology, Karachi Pakistan. He completed his PhD from the University of Canterbury, New Zealand, in 2015. He has been holder of REANNZ PlanetLab New Zealand scholarship and Battersby Trimble award for advancement of computing in New Zealand. As Principal Investigator Asad has won various competitive research grants from the Higher Education Commission of Pakistan including HEC NRPU and HEC Technology Transfer Support Fund. He has also won a 118 million PKR competitive research grant for establishing the National Research Centre for Cyber Security in Pakistan. He is also currently heading the IT department and Network Operations Centre of NED University of Engineering \& Technology, Karachi, Pakistan. Dr. Asad has extensive research collaborations with the WAND research group of the University of Waikato, New Zealand, and the DFKI German research centre for Artificial Intelligence. On industrial front, Dr. Asad is developing SIEM solution for National Clearing Company of Pakistan (NCCPL) with TTSF grant.


(HEC NRPU Project by PI Dr. Asad Arfeen)


Traditionally the Internet traffic was dominated by client server communication model. This means that Internet is designed to be content-retrieval network where end hosts act as information-sink only. With centralized content and distributed clients at the edge of Internet, the traffic pattern was highly asymmetric with downlink (carrying content) overwhelmingly dominating the uplink utilization (carrying content requests only). With the advent of peer to peer applications and on-line social network applications (for example, Facebook, Twitter, Whatsapp), the trend saw a significant change in late 2000s. With the rapid advancements in Internet access technologies supporting high bandwidths, popularity of social media and reduced smart phone prices with high resolution cameras, the content has been commoditized and dispersed.
The emerging and future paradigm of the usage of Internet with new applications and services will break Internet traffic asymmetry. That is end users are no longer acting as information consumers or sink only. They now generate significant amount of information content which is both temporal and non-temporal in nature. Therefore, traffic emitting from access networks (upload) is expected to become equal to or even exceed the volume of traffic sinking (download) in access networks. Hence there is a great need to revisit traditional models for Internet traffic. Internet traffic streams can be highly correlated stochastic processes with merging and splitting occurring at various points. Therefore, understanding and modeling these processes is important in the case of new emerging paradigm of Internet traffic.
Once the models of Internet traffic are developed, it will be very important to develop efficient estimation methodologies to evaluate parameters of the models. The issue becomes more challenging because of the size and speed of huge data sets. Any attempt for offline evaluation will require large storage and computation capabilities. Hence the main task here is to develop statistically sequential methodologies which do not require large samples for evaluation. The sequential algorithms will workon large traffic data sets in a recursive fashion and will keep track of the state and health of current Interent traffic. These sequential algorithms can also be used in Internet traffic prediction or forecasting tasks which can help in appropriate network resource allocation and bandwidth provisioning tasks.
The outcomes of the proposed project will establish Pakistan's position in global effort for Future Internet (FI) design. Pakistan's Internet service provider industry and traffic regulating authority will greatly benefit from the planned theoretical and practical work under the proposed project. The Internet traffic archives will help them in various applications, for example, detecting intrusions, understanding traffic congestion events, predicting traffic profiles and understanding current and expected application mixture.