The programme-wide research is centered on the use of adaptivity in computing. Future computers will no longer be centrally located in boxes as we currently know it but will instead be pervasively embedded into the environment. For such computer systems to work effectively, many technical breakthroughs shall be necessary. The research in the SMA’s CS programme is targeted towards solving two of the main problems in devising such systems.

High Performance Computational Cluster - CS

The first main research problem lies in having to make the system context aware. The computer will not be able to assist humans effectively unless it is aware of all the important factors that currently do or have the potential to affect the outcome of any action. Building computer systems which are aware of all the important factors in the environment poses a technical dilemma, because characterising the environment and behaviour may involve very complicated sets of parameters which not only change, but also change rapidly. Research on working solutions to such problems is the main theme of the Adaptive Intelligence research thrust in CS. The second research problem lies in creating an autonomic system, that is enabling it to be aware of its own resources and be able to take appropriate actions to guarantee levels of service in the presence of failures in its components. This is again a very challenging problem as the scale and complexity of systems are both growing fast beyond human capability. Research on such problems is the main theme of our Adaptive Software research thrust.

To tie the two thrusts together, research has been focused on components that can support a useful application. Initiatives have been made for collaborations with the Institute for Infocomm Research (I2R) to work on an adaptive computing system that is able to help the elderly to live independently at home. Such a system would be easy to set up (self-configurable), able to work well in the presence of partial failures (robust) and be easily connected together into a grid for information and resource sharing. It will be able to understand the environment well enough to send alerts to caregivers of the elderly in cases of emergencies such as falls, strokes or heart attacks. It will be able to alert the elderly in case they miss scheduled events, such as the consumption of medication. It will also be able to summarise the activities of the elderly to allow caregivers to easily query and monitor the sleep, exercise, eating and other important activity patterns of the elderly. Labs have been set up in NUS, NTU and I2R, equipped with mobile robots, cameras, microphones, wearable sensors, projectors and various pervasive computing devices in order to perform the research. Several Ph.D. students are currently working on these topics.