Asynchronous Design Methods
Between the hammer of the continued semiconductor scaling and increasing degree of integration on the one side and the anvil of increasing limitations on and unpredictability of power and energy, computation systems will require more and more timing flexibility and heterogeneity even within individual chips.
The µSystems group has a long history of successful research from theoretical methods to practical implementation on all matters related to system timing and asynchrony, and is continuing with a number of research projects on synchronization, SoCs and NoCs, GALS and asynchronous systems.
Current and future ICT systems are and will more and more be bound in their capabilities and operations by the energy available to them. Energy driven computing is a new paradigm which states that computation should be a result of energy supply (computing on energy).
The µSystems group is one of the initiators of this paradigm and is detailed in Microelectronics for autonomy and survival (PDF: 175KB) about the applications of this type of computing and the methods with which it can be designed.
Systems for survival
With energy and power becoming more scarce and their sources more unreliable and unpredictable, especially for certain embedded systems based on harvested energy mostly or even solely, how to best deliver such energy to computation units and how such computation units can be designed to best make use of this energy become critical research problems.
A number of our current projects try to answer these questions by increasing the system survival zone, creating methods for near-threshold computing system design, enhancing the variation tolerance of circuits, and developing efficient on-chip power delivery and conditioning including control and sensing techniques.
Power-performance-reliability interplay in many-core systems
Power and energy availability not only affects system performance and survivability, it also impacts system reliability. There exists a complex, but poorly understood, interplay among power supply (related to voltage), system computation throughput, reliability, and how fast the circuits age and degrade.
The µSystems group is applying its considerable expertise in related areas to new research which aims to improve the understanding of these complex relationships and develop methods to improve system designs based on such discoveries.
Modelling and design methodologies for extremely low power
The µSystems group is traditionally very closely connected to the Computing Science School of Newcastle University and continues to carry out research on the edge between computer science and electronic engineering.
Currently we are engaged in research on powerful modelling methods to support the design of computation units including microprocessors, for environments where power and energy supplies are uncertain.
This research is led by Andrey Mokhov.
One crucial technological area where computing and electronics could generate considerable new ways of helping society is the potential use of computing in real-time health monitoring and care. An exciting aspect in this is the interfacing of silicon-based computation devices with biological neurons.
System design for optogenetic retinal prosthesis
The µSystems group includes the Neuro-prosthesis lab, with a primary interest in developing neural stimulators and state of the art implantable systems for the new field of optogenetic neuroprosthesis. Along the way, we hope to both generate new understanding in core neurobiology and use neuroinspired designs to make better circuits and systems.
For more information please see the Neuro-prosthesis lab.
Telemetry for Neuroscience Research
A collaborative project between the µSystems group and the Institute of Neuroscience aims to provide an enabling technology to support neuroscientists in their research into how the brain works. In the long term, this research will lead to improved treatment of nervous system injuries such as strokes and spinal cord injuries.
A radio telemetry system is being developed to transmit multiple muscle or nerve signals from a device which is implanted within the body. Power is supplied by radio frequency induction, from an external coil placed over the skin, avoiding the need for battery replacement.