No abstract

We present E 2 RINA, an aggregation algorithm for wireless sensor network applications characterized by clustered topologies, such as building automation and manufacturing plants. Thank to an efficient use of the wireless channel, E 2 RINA offers the robustness of the gossip-based algorithms and, at the same time, the energy performance of the faster cluster headbased algorithms. We also developed a mathematical model to predict the performance of the algorithm with respect to the free variables without the need of extensive simulations. We validate our model and the robustness of E 2 RINA by running a simulation model of a test case consisting of a cluster of MICA nodes.

No abstract

The continuous shrinking of process geometries increases variability and demands for conservative margins that have a negative impact on performance. With conventional clocks, the cycle period has to be defined to accommodate the worst-case variations during the lifetime of the circuit. Elastic Clocks arise as a new paradigm to reduce the margins without sacrificing robustness. Their cycle-by-cycle adaptation to static and dynamic variability enables the use of reduced margins that only need to cover the differential variability of the circuit delays with regard to the elastic period. Given the substantial spatio-temporal correlation within every die, a significant reduction in the margins required to cover process variability, voltage and temperature fluctuations and aging can be achieved.