PhD Jawwad Ahmed

– I am originally from Multan in Pakistan. I got a Bachelor degree in Computer Science from Punjab University in Lahore. I received my Master degree in Information Technology with a major in Network Technologies from National University of Science and Technology (NUST) in Rawalpindi, Pakistan. After acquiring my Master’s degree I also spent some time in the software development Industry in Pakistan. In 2008, I got a position as a PhD student at KTH and joined soon afterwards.  

– My research topic is related to Optical Networking for my PhD studies. My thesis project for the Master’s degree was in the area of Optical Networking as well. During that time I started to develop an interest in the field as I became more familiar with the topic. Later on I was happy to discover that there is indeed a very active group working in a similar research field at KTH. Since I already gained knowledge in this area from my Master thesis project I decided to continue along the same line and pursue my PhD studies in this domain. 

– Optical networks based on Wavelength Division Multiplexing (WDM) technology possess many significant benefits in terms of high capacity, flexibility and low power consumption. All these advantages make WDM networks the preferred choice for todays and future transport solutions which are strongly driven by a plethora of emerging online services. In WDM networks, the capability to provide high capacity during the service provisioning phase is very important, together with traffic dynamicity which is another essential aspect to consider because in many scenarios, for example in the case of real time multimedia services, the connections are expected to be provisioned and torn down quickly and relatively frequently. Furthermore, survivability, in the presence of new failure scenarios is critical to minimize the network disruptions and data losses. Furthermore, service provisioning is also important in the access paradigm where Long reach Passive Optical Networks (PONs) are gaining popularity due to their cost, reach, and bandwidth advantages. Long Reach PONs (LR-PONs) introduces additional challenges in terms of network performance in dynamic provisioning scenario. It becomes quite apparent that effective solutions to the connection provisioning problem in both the core and access optical networks with respect to the considerations made above can ensure a truly optimal end-to-end connectivity while making an efficient usage of resources.

– The work in the thesis addressed as the dynamic resource provisioning problem in optical core and access networks based on the wavelength division multiplexing (WDM) and Passive Optical Network (PON) paradigm, respectively. Furthermore, in optical core networks, the problem of survivable provisioning was also taken into account, considering both single and double link failure (DLF) scenarios.

With respect to the work related to WDM-based core networks, we proposed a centralized bulk provisioning framework with the objective to enable the concurrent provisioning of multiple connection requests using the Path Computation Element (PCE). An Integer Linear Programming (ILP) model was also developed to optimally solve the problem of concurrent provisioning of connection requests at the PCE in a dynamic traffic scenario. To minimize the impact of the path computation delay on the connection setup times, a Greedy Randomized Adaptive Search Procedure (GRASP) meta-heuristic was also proposed. Our simulation results illustrate significant performance gains that can be achieved by the proposed bulk provisioning framework in terms of blocking performance, efficient resource utilization, and control overhead reduction in PCE-based WDM core networks as compared to provisioning schemes which provision connection in strictly serialized manner without concurrent optimization. 

Considering the importance of resilience in optical networks, we extended the proposed bulk provisioning framework to account also for connection requests requiring path protection against single link failures. In particular, we proposed a GRASP-based meta-heuristic to enable Shared Path Protection (SPP) based provisioning. In addition, we also developed a simpler (i.e., greedy) heuristic used for concurrent provisioning of connection requests. All the proposed survivable approaches can be readily deployed in the dynamic bulk provisioning framework described above. Simulation results confirmed tangible gains in terms of blocking performance, primary/backup resource utilization, and high backup resource shareability in case of SPP, when compared to a sequential approach.

Protection against single failures is critical, but the ability to guarantee high service availability, in the presence of multiple failures cannot be ignored either. With this objective in mind, we proposed two Double Link Failure (DLF) recovery schemes that focus on minimizing the connection disruption period. The performance evaluation showed that with the proposed provisioning approaches, the average connection downtime as well as the number of dropped connections can be significantly reduced. This was achieved while utilizing much less backup resources in comparison to conventional path protection based techniques used to mitigate the impact of double link failures.

Our research contributions to the optical access networks addressed the Dynamic Bandwidth Allocation (DBA) perf ormance degradation issues in Long Reach Passive Optical Networks (LR-PON). We proposed two novel schemes, namely Newly Arrived Plus (NA+) and Enhanced Inter-thread Scheduling (EIS). Both schemes utilize a multi-threading mechanism to mitigate the impact of excessive grant delays that occurs in LR-PON due to the reach extension. Performance results conducted for different load and reach scenarios confirm that EIS can significantly improve packet delay performance as compared to the benchmark scheme. Furthermore, jitter and network throughput performance is also significantly enhanced over the compared multi-threading scheme.

– In terms of future work, in the area of core networks it might be interesting to assess the maximal performance gains achievable by employing an optimal approach at the PCE for shared-path-protection (SPP) based dynamic bulk provisioning. It is also worth to explore the advantages of dynamic bulk provisioning when applied to packet switching layer in a multilayer networking scenario with (or without) deploying it at the optical layer. The reason is that packet switching layer typically has much higher traffic dynamicity than optical layer in current transport networks. Furthermore, it might be interesting to explore the benefits of dynamic bulk provisioning when applied in a grid (or cloud) computing environment where joint optimization of network and grid resources can result in significant performance benefits.

An interesting future work for the DBA design in LR-PON is to devise an automated mechanism to dynamically adjust/tune the number of employed threads in the proposed schemes based on the current load and reach values offering optimal performance under a specific network scenario. Furthermore, proposed DBA schemes should be extended to support differentiated service requirement scenarios based on tolerable delay threshold for each traffic class.

For now the plan is to stay at KTH for a few months to continue my research in the area of optical networking. There are still some research topics in this domain I am currently pursuing and I would like to wrap-up my research on these topics after my PhD. Later on, maybe I will try to find a suitable job preferably in the same or a similar research domain in Sweden either in Academia or industry. So, for now the plan is to stay here in Sweden, and get some more work experience in a research environment that will be valuable later on.

The thesis by Jawwad Ahmed is to be found here .