Research

Research Groups > Systems

Gene Cooperman Gene Cooperman   Peter Desnoyers Peter Desnoyers   Alan Mislove Alan Mislove   Guevara Noubir Guevara Noubir   Ravi Sundaram Ravi Sundaram

Northeastern's systems group focuses on the middle layer between hardware and software, seeking to take advantage of efficiencies possible with the new architecture of computers.

The architecture of personal computers has changed significantly in recent years. Individual sequential programs aren't running faster, but it is now possible to efficiently run four program threads at once. That number may get much larger. By the year 2011, 80 processors on a single chip may be possible. At the same time, memory system bandwidth and size show no sign of scaling to match. New approaches are required to extract performance from these highly multi-core CPU architectures of the future.

Collaborating with CERN (European Organization for Nuclear Research) to take advantage of the coming many-core systems, the systems group is working on a way to automatically convert multiple large processes (of over a million lines of code) to one process with multiple threads using shared data, to create a smaller memory footprint.

The security and performance impact of multi-core processors on cloud computing is another major research focus. The group explores attacks on multi-core schedulers as well as techniques for disruption-proof migration of virtual machines over a networked heterogenous substrate of multi-core processors.

The system group also focuses on the area of emerging solid state storage systems. For four decades, computer data has been organized in ways dictated by the performance limitations of disk drives. Solid-state drives (SSDs) and embedded flash enable the development of new ways of handling data.

In networking, the group works on the design and prototyping of scalable, robust, and secure wireless communication systems. Areas of focus include cross-layer cooperative communication mechanisms and the resiliency to significantly reduce energy consumption and increase network capacity. These systems make use of distributed cross-layer diversity techniques that combine long-range cellular links (e.g., 3G+), short-range high-speed links (e.g., WiFi), and low-power, low-speed links (e.g., ZigBee).