Research
Primary Research Interests
- Heterogeneous Wireless Networks
- Wireless Mesh Networks
- Pervasive Computing & Autonomous Networking
- Scheduling & Resource Allocation Techniques
- STDMA Scheduling
- Mixed-Biased Scheduling & Resource Allocation
- Load Balancing
- Cross-Layer Design & Optimization
- Simulation [ns2, ns3]
- Mathematical Models for Networks
- Cognitive Radio Technique
- Heuristic Approaches
Other Research Interests
- Operating Systems
- Distributed and Parallel Systems
- Reconfigurable Computing
- Field Programmable Gate Arrays (FPGAs)
- Configurable Interconnection Networks (CRINs)
- Soft Computing, Artifical Intelligence
Research Presentations
Heterogeneous Wireless Networks
Heterogeneous Wireless Networks have recently become an area of interest to me. Since there is already a plethora of existing radio access technologies (RATs) it makes sense to try to take advantage of the existing technology. Increasingly, user devices are equipped with multiple radios which makes this type of work attractive. The user device can then either select the best available access technology, or in some cases even use multiple access technologies at once in order to increase the amount of network capacity available to the device. The device selects the access technology based on a range of parameters such as the cost of access on a pariticular network, the capacity of the network and the demand at the user device, the quality of the wireless links and so forth. This type of wireless networking will increasingly lead to more pervasive networks, allowing people to connected anywhere they go with broadband network access. Wireless network providers will be able to increase capacity in congested areas of networks by partnering with smaller scale providers using alternate technologies such as 802.11 Wifi or even deploy their own small scale wireless networks as an altnerative to expensive nano and pico cells.
Wireless Mesh Networks
Primarily, I am interested in research in wireless networks, specifically Wireless Mesh Networks (WMN). I feel this is an important area to study because compared with wireless local area networks (WLANs) a wireless mesh network supports multiple base stations or access points for wireless clients to connect to. There is also support for multiple redudant paths which enables load balancing and scheduling to be applied within the network. This property also makes the network more reliable because several of the base stations could leave the network and there could still be communication whereas in a single AP situation such as WLAN this is not the case.
When compared with ad-hoc networks, a WMN has the advantage that the BS nodes have more resources available to them compared with an ad-hoc node. These resources could be battery power (or signal power), processing resources or memory since it is generally assumed that these nodes form the infrastructure of the network and are connected to power. In contrast an ad-hoc node is usually mobile and constrained by battery power limitations. Additionally, each node in an ad-hoc network performs routing whereas in a WMN the mesh routers perform most of it. Another interesting obvservation about WMNs is that generally the traffic flows from gateways (GWs) to clients and from clients to GWs. In an ad-hoc network this is not the case. This assumption allows us to tailor algorithms specifically to this task since the primary goal of a WMN is to distribute a connection to the internet to many clients over a large area.
Pervasive Computing & Autonomous Networking
Pervasive computing is the ability to compute anywhere, and the strong integration of computational devices and technology into our lives. This should be achieved in such a way that it is invisible to the end users. When we apply this concept along with computer networks, the possibilies increase. With the latest advances in wireless technologies, there are many problems with mobility, interference management of the network. Increasingly, it is becoming difficult and complex for human administrators to manage these networks. Autonomous networking helps to create a pervasive computing environment by keeping the network invisible to end user. Devices should be able to join and leave the network freely, between various technologies without management. This introduces many problems which are often difficult to solve such as mobility, handoff / handover, interference, gateway placement between various technologies.
Scheduling & Resource Allocation Techniques
The main motivation for studying fair scheduling in WMNs is commercialization of these networks. If these networks are ever to become commercially viable, one large challenge that must be overcome is how to ensure that all users in the network receive equal Quality of Service (QoS). When each user pays the same amount of money to access the network, each expects similar service quality however in many implemenations this is not the case. A greedy user may monopolize all of the network resources so that other legitimate users are penalized. The network must support fair scheduling while still maintaining good throughput and finding a balance between these two problems is the key to this research.
© Jason Ernst 2008, University of Guelph
Cross-Layer Design & Optimization
Cross-Layering has become popular in recent years in Wireless Network research for one main reason. This is to increase the performance of the network. Cross-Layering can be used to share information from layers which do not normally communicate in the standard layered model of networked communication. This sharing of information can enable informed decision making that can greatly increase the performance of the network. For example, information on the quality of the links in the network can be considered when making decisions on routing so that only the best quality links are used to ensure that packets are received without errors more often. Cross-Layered design must be carefully planned however, in order to avoid problems with extending the work in the future as it can lead to unmaintainable "spaghetti" code if not carefully designed. This approach can almost be seen as a trade-off in performance versus maintainability.
Page last Updated: 11/19/10
