Research Overview

Research Interests

• Artificial Intelligence

• Internet of Things, Wireless Sensor Networks

• Quality of Experience (QoE), Quality of Service (QoS)

• Network Function Virtualization/ Software Defined Networking

My work focuses on AI and optimisation approaches for enhancing networking performance. My different contributions are grouped and organised into 3 main parts. The research direction is towards designing intelligent networks. The first part uses ML based approaches for detection and estimation problems in the network. The second part uses the detection and estimation results with optimisation techniques to improve routing and virtual network function placement methods. The last part is about improving AI and data analyses algorithms and integrating them further to go towards automation as well as autonomous networks in future.

1. Estimation and detection

How can we automatically detect Quality of Experience (QoE)? More generally how can we detect user context, which is linked with QoE? Can ML detection be done in-network or in data plane itself?

We contribute by providing QoE estimation methods, user context detection and in-network detection using ML. The work on QoE-estimation was done during my postdoc stay at INRIA Rennes. ML was used for perceptual quality estimation of video streaming applications. The work related to user context detection was done in the PhD thesis of Illyyne Saffar, which I co-supervised.

In the context of estimation and detection, we also show that in-network detection is also possible by embedding ML inside network data planes. We propose SwitchTree which embeds the Random Forest algorithm inside a programmable switch such that Random Forest is configurable and reconfigurable at runtime. We show how some flow level stateful features can be estimated, such as the round trip time and bitrate of each flow.

For example, please see :

Illyyne Saffar, Marie-Line Alberi-Morel, Kamal Deep Singh, César Viho. Deep Learning based Speed Profiling for Mobile Users in 5G Cellular Networks. IEEE Globecom, December 2019, Hawaii, USA

Kamal Deep Singh, Yassine Hadjadj-Aoul and Gerardo Rubino." Quality of Experience estimation for adaptive HTTP/TCP video streaming using H.264/AVC ". CCNC 2012, Las Vegas, Nevada, USA, Jan 2012.

Jong Hyouk Lee, Kamal Singh. "SwitchTree: In-network Computing and Traffic Analyses with Random Forests", Neural Computing and Applications, 2020

2. Network Optimisation

Can we use the information obtained through estimation and detection to optimise the performance of the network?

We contribute by proposing centralised as well as distributed QoE-aware optimisation algorithms.

After estimating QoE and user context, the next step is to use this knowledge for network optimisation and improving user experience. We focus on QoE-aware optimisation using the QoE estimation tool and optimization techniques in the phd thesis of Quang Tran Anh Pham . Additionally, network virtualisation techniques have recently enabled flexibility and we exploit this flexibility to optimise the network using QoS-aware placement of virtual network functions (NFV) which will be extended to QoE-aware placement in future.

For some works please see:

A Najjar, Y Mualla, Kamal Singh, G Picard, D Calvaresi, A Malhi, S Galland. ``One-to-Many Negotiation QoE Management Mechanism for End-User Satisfaction'', IEEE Access, 2021

Tran Anh Quang Pham, Abbas Bradai, Kamal Deep Singh, Gauthier Picard, Roberto Riggio. “Single and Multi-domain Adaptive Allocation Algorithms for VNF Forwarding Graph Embedding”, IEEE Transactions on Network and Service Management, Volume 16 , Issue: 1 , March 2019.

Tran Anh Quang Pham, Kandaraj Piamrat, Kamal Deep Singh, César Viho. "Video Streaming over Ad-hoc Networks: a QoE-based Optimal Routing Solution". IEEE Transactions on Vehicular Technology, 66(2): 1533-1546, 2017.