S. (Saad) Saleh, MSc

Cognitive In-network Processing using Software-defined Communication Architectures

In the Internet, many applications are highly distributed and they typically require low latency in processing data streams. In addition, communication networks need to be adapted to process packets in the context of highly dynamically changing environments. "Cognitive packet processing " has the potential to enhance network protocols to adapt to a variety of different situations and can help to perform, in the context of data-intensive applications, the processing on the path between producers and consumers of information. This project studies the implementation of novel emerging hardware technology "Memristors" for providing computing and cognition in network architectures. The contributions are three-fold,

(1) enhancement of software-defined networking architectures so they can perform cognitive in-network processing.

(2) understanding of performance trade-offs and cost of architectures in controlling and executing cognitive in-network processing functions.

(3) understanding of how cognitive in-network processing can enhance communication protocols at different layers of a communication system, e.g. transport or application layer.

Covert link detection in Instant Messaging Networks using Flow log data

Studying an attack on relayed instant messaging (IM) traffic that allows an attacker to infer who's talking to whom with high accuracy. This attack only requires collection of packet header traces between users and IM servers for a short time period, where each packet in the trace goes from a user to an IM server or vice versa. The packet header traces contain information such as timestamps, IP addresses, and port numbers of users and IM servers. Note that packet payloads are encrypted by most IM services; therefore, they cannot be used to infer which users are talking to each other. The specific goal of this attack is to accurately identify a candidate set of top-k users with whom a given user possibly talked to, while using only the information available in packet header traces. It is technically challenging to perform this attack in real-world IM networks due to simultaneous one-to-many communication among users, unequal transmission delays, and out of order or duplicate packets.

On the Quality of Service (QoS) of IPTV and VoIP over IEEE 802.11n

Internet Protocol Television (IPTV) and Voice over Internet Protocol (VoIP) have gained unprecedented growth rates in the past few years. Data rate and high coverage area of IEEE 802.11n motivate the concept of combined IPTV and VoIP over IEEE 802.11n. Transmission of combined IPTV and VoIP over a wireless network is a challenging task. In this research, we deal with the capacity evaluation of combined IPTV and VoIP over IEEE 802.11n. We evaluate the use of Datagram Congestion Control Protocol (DCCP) at transport layer of IPTV and VoIP. We propose a multicast congestion control protocol for efficient data delivery with better quality of service.

A Zero-input Cloud-based Application to Measure and Map Road Conditions in Pakistan

We propose to design and develop a smartphone app that detects when it finds itself on a road and measures shocks and vibrations in order to A) map and measure the state of disrepair of roads, B) map the locations of potholes and C) map the locations of speed bumps. The app will not require any input from its user and report measurements to a cloud hosted application which stores data from cellphones and jointly processes the data from all users to obtain a better and more complete estimate of road conditions. This map will be made publicly available via a website that will allow citizens and municipal authorities alike to spot potholes and road segments in need of repair, as well as imbalances in infrastructure maintenance efforts across cities. If the municipality has data of approved and legally constructed speed bumps, it is able to identify all others as illegally constructed.

VNC: Vehicular Network Coverage for Multihop Quality-of-Service Data Delivery

Working over a Vehicular Network Coverage (VNC) algorithm for providing vehicles with the Quality-of-Service (QoS) Data Delivery in vehicular networks using infrastructure nodes, such as Road-Side Units (RSUs) and Relay Nodes. Two multi-hop data delivery directions have been considered i.e. outbound delivery from vehicle to RSU and inbound delivery from RSU to vehicle. We define the QoS of data delivery with the mean and standard deviation of End-to-End data delivery delay. For the given requirements of both outbound and inbound QoS delivery, we need to deploy an appropriate number of RSUs. In this research, we propose the optimal deployment of RSUs.

Predicting New Collaborations in Academic Citation Networks of IEEE and ACM Conferences

Of key interest in the study of social networks is their evolution with time. Among social networks, collaboration networks have been the subject of significant interest with regards to link prediction. The accurate prediction of the appearance of new collaborations between members of a collaboration network can help accelerate the realization of new synergies, foster innovation and raise productivity. For this study, the authors collect a large data set of publications in 630 conferences of the IEEE and ACM containing more than 257,000 authors, 61,000 papers, capturing more than 818,000 collaborations spanning a period of 10 years. The data set is rich in semantic data that allows the exploration of many features that were not considered in previous approaches.

Post Summarization of Micro-Blogs of Sporting Events