Non-linear energy harvesting and approximate computation for energy-autonomous low-power devices

Wireless sensor networks are technologically mature and have attractive applications, but are not widely used, mainly due to energy constraints.

Energy efficiency in wireless sensor networks has been explored mainly at the level of communication, developing ad-hoc protocols and routing mechanisms, but to ensure good results in terms of energy efficiency an effort at system-level is needed.

One of the key points that play a major role on wireless sensor networks lifetime is the power supply of the nodes. In this seminar will be presented the feasibility of powering the wireless nodes using the energy present in the environment, focusing on energy present in the form of vibrations. Since the environmental vibrations are usually random, and then spread on a wide spectrum of frequencies, methods of energy-harvesting based on nonlinear dynamics, that have shown outperform classical methods for the conversion of this type of vibration, will be presented.

On the other hand accepting compromises on computation result accuracy the energy consumed can be drastically reduced. In the second part of the seminar the fundamentals limits of energy consumption for computation will be presented, correlating the minimum energy required for calculation and resulting accuracy.

Rapporteur

Igor Neri - University of Perugia

Teacher reference

Alessandro Bogliolo

Participation constraints

Nessuno

Dates

Place	Date	Time	CFU
Aula Turing	March 11th, 2014	16:00-18:00	0.125