Smart Applications on Virtual Infrastructures
Alberto Leon-Garcia is the Scientific Director of the NSERC Strategic Research Network for Smart Applications on Virtual Infrastructures (SAVI). SAVI investigates the development of an extended computing cloud that encompasses remote datacenters as well as a “smart network edge” to support services platforms where new applications can be readily deployed, scaled up and down, and retired. SAVI investigates the role of virtualization to develop enabling software and networking technologies and a testbed to explore Future Internet protocols, network-based cloud computing services, and large-scale applications. SAVI involves researchers from nine universities as well as partners from industry and government.
To see publications pertaining to SAVI click here.
To find out more about SAVI click here.
Connected Vehicles and Smart Transportation
Prof. Leon-Garcia is the Principal Investigator for this multidisciplinary ORF Research Excellence project. The objective is to create an open and flexible applications platform for Connected Vehicles and Smart Transportation (CVST) systems that leverages emerging and future onboard sensor, wireless communications, and cloud computing technologies. This platform will integrate vehicles, interconnected by vehicular and wireless communications to transportation infrastructure and to the Internet. The platform will also collect data about traffic and road conditions from road sensor systems as well as new data that will be generated by sensors in connected vehicles, including mobile devices, cars, buses, trucks, and travellers. This richer and higher quality data will be used by the private sector to offer novel Internet applications and by publicly operated smart transportation systems that regulate traffic in streets, roads, and highways. The proposed shared approach to gathering and distributing data is expected to reduce the cost of the required infrastructure.
To see publications pertaining to CVST click here.
We have developed a new metric, network criticality, to characterize the robustness of a network. We represent a network as a weighted graph where the weights are a measure of capacity. We show that each weighted graph has a “criticality” value that depends on the topology and weight matrix of the graph. A low value of criticality indicates robustness with respect to changes in network flows, link capacities, or topology. The network criticality is a convex function of the weights and flows so a variety of optimization problems can be posed to maximize robustness. In particular, the optimality conditions provide a means to derive autonomic control laws and traffic engineering methods to assign resources subject to unforeseen changes in topology and traffic demands. We have applied the method in flow assignment and fault management, as well as load balancing and topology design. In the CVST project we are exploring the application of network criticality to transportation networks.
To see publications by Prof. Leon-Garcia on Network Criticality click here.
Green Networking & Cloud Computing
Information and communications technology can play a key role in the reduction of green house gases by enabling smart resource management systems that can regulate the usage of resources in large scale systems, including cloud computing, networking, transportation, and power grids. We are exploring the application of network criticality in the design of green networks that promote the use of renewable energy in network infrastructure. This work is part of the GreenTouch consortium. We are also involved in a project addressing the design of green telecom service provider infrastructure.
A video of a talk by Prof. Leon-Garcia on smart infrastructure is available here; publications here.
The future smart grid will require the integration of plug-in hybrid electric vehicles, massive energy storage and distributed renewable energy sources. A key element is the development of new management algorithms that match demand and supply, improve performance, stability and security, and offer cost-effective and environmentally friendly solutions. The deployment of sensors and communication technology in the grid, in vehicles, and in homes and buildings leads to a more dynamic and distributed management infrastructure. Market-based algorithms, which provide incentives to the producers and consumers of electricity as well as intermediaries such as transmission and storage facilities, have the potential to provide effective management solutions. We are developing a market-based approach, posed under a graph-theoretic framework, that addresses the challenging management problems of the future smart grid. These include the integration of hybrid vehicles, distributed storage technology, renewable energy sources, cloud computing that is closely coupled to the grid, and demand response systems. The work considers system security and stability, metrics and utility functions to assess performance, robustness, as well as financial and environmental cost.
To see publications by Prof. Leon-Garcia on Smart Grids click here.
Optical Fabrics for Datacenters
In this CIPI project we consider the design of optical networks that can provide connectivity for future datacenters that can interconnect several million servers. At this scale, power consumption and space for Ethernet cabling become severe challenges. We are designing optical switching fabrics that combine WDM and burst switching to provide extremely high capacities with high spatial compactness and low power consumption. This joint work with partners at McGill and Laval Universities considers switch architecture performance as well as optical impairments effects. In related work with colleagues from the Computer Engineering Group at U of T we are considering the design of optical interconnect fabrics for future multicore chips.
For publications by Prof. Leon-Garcia on Optical Fabrics click here.
Autonomic Service Architecture
As Canada Research Chair in Autonomic Service Architecture (ASA) since 2005, Prof. Leon-Garcia is developing the architecture for new service management and control systems that largely manage and control themselves, and are able to accommodate a multitude of existing and future applications, thus promising to be highly cost efficient and flexible. The aim is to design a self-regulating management and control system that is responsive to ever changing demands as well as to equipment failures. Such systems will autonomously regulate and optimize configurations of data flow, be able to protect themselves from harmful impact and have the capabilities to self-heal. The ASA project led to the following research outcomes:
- An Application-Oriented Network Architecture (AON) that uses service-oriented approaches to provide virtual networks of computing and communications resources to support large-scale applications. The VANI project (see below) developed a prototype implementation of AON.
- Autonomic mechanisms for regulating the resources allocated to virtual networks to support changing demand and to deal with failures. These mechanisms are based on the Network Criticality metric developed by Ali Tizghadam and Prof. Leon-Garcia (discussed below).
To see publications by Prof. Leon-Garcia related to ASA click here.
Virtual Application Networking Infrastructure
The Virtual Application Networking Infrastructure (VANI) is our effort to develop a converged computing and communications infrastructure to support an open applications marketplace using our AON architecture. VANI nodes offer virtualized processor, programmable hardware, networking, and storage resources that can be interconnected to provide virtual networks to support new protocols and applications at scale. The VANI node is the starting point for the national application testbed that is being developed as part of the SAVI NSERC Strategic Network. To learn more about VANI click here.
To see publications by Prof. Leon-Garcia on VANI, click here.