Research departments


OMNI TEAM: Optimisation, Mobilité & Réseau

(Optimization, Mobility, Networking)

Team leader

Eugen DEDU


Microelectromechanical systems (MEMS) are now sufficiently well developed to be mass produced and incorporated into products for everyday consumption. Gyroscopes and accelerometers are henceforth included in the ABS systems of automobiles, as they are in the latest cell phones and in the matrices of the digital micro-mirror devices (DMD) now found in most video projectors. These MEMS may be used alone, as is the case in accelerometers, or they can be grouped to work together towards a common goal as is the case of DMDs or distributed MEMS.

A MEMS may also have an integrated data processing capability. It may be centralized on a PC or an FPGA, but extensibility may then be a problem which can be solved by a distributed architecture; in this case it is called an intelligent distributed MEMS (DiMEMS).

omni blocksimulator
Goals and Research Areas

A DiMEMS system is an aggregate of entities all of which include a sensor and/or an actuator, a computing unit and wired or wireless means of communication. DiMEMS are classed according to their characteristics: type of MEMS (sensor, actuator, sensor/actuator) which sets the information flow, type of network (wired or wireless) which ensures transmission reliability, network topology (static or dynamic) and the degree of actuator synchronization, if any (none, local, global). The combination of characteristics will define the different type of DiMEMS which will potentially require different information management strategies.

Indeed, a DiMEMS that includes sensors, a wired network and a static typology will not be subject to the same constraints and scientific locks as a DiMEMS composed of sensors/actuators, communicating in an unreliable network and that is mobile.

However, all DiMEMS have properties in common, different from other computing systems:

  1. The first property is extensibility management. Indeed, like MEMS, such systems are mass produced and a great many can be used for a modest price. All algorithms must therefore take their use into account and anticipate a scalability that reaches into the millions.
  2. The second is management of communication density. One must imagine that within the volume of a single cubic meter of water it is possible to have as many links and entities as exist on the entire public internet. This density of communication must therefore be managed, especially in the case of wireless communications.
  3. The third property is that it is not yet possible to dispose of an equivalent computing capacity between the macro and micro worlds. A single unit of computing in an intelligent distributed MEMS is, at most, on the order of one square millimeter. Computing must thus be adapted accordingly.
  4. Lastly, all of these systems interact with the outside world via sensors and/or actuators, which poses problems of real-time and of synchronization.

The team considers DiMEMS to be a subdomain of distributed computing and more particularly, where interaction with the real world is concerned, of the Internet of Things (IoT).


The scientific tools used by the OMNI team are numerous. Simulation and emulation of DiMEMS systems play an important role as they support the development of specific distributed algorithms within DiMEMS such as the distributed reconfiguration of modular robots. The multiscale positioning and mobility are intended to provide new positioning methods, from the metric to the micrometer scale.

A unique expertise has also been developed surrounding modeling, simulating and optimizing wireless networks within the 1-10 GHz band frequency, thus allowing for wireless communications in DiMEMS, but also between different processor cores.