Section: Scientific Foundations
Interaction between Algorithms and Architectures
As Cairn mainly targets domain-specific system-on-chip including reconfigurable capabilities, algorithmic-level optimizations have a great potential on the efficiency of the overall system. Based on the skills and experiences in “signal processing and communications” of some Cairn 's members, we conduct research on algorithmic optimization techniques under two main constraints: energy consumption and computation accuracy; and for two main application domains: fourth-generation (4G) mobile communications and wireless sensor networks (WSN). These application domains are very conducive to our research activities. The high complexity of the first one and the stringent power constraint of the second one, require the design of specific high-performance and energy-efficient SoCs. We also consider other applications such as video or bioinformatics, but this short state-of-the-art will be limited to wireless applications.
The radio in both transmit and receive modes consumes the bulk of the total power consumption of the system. Therefore, protocol optimization is one of the main sources of significant energy reduction to be able to achieve self-powered autonomous systems. Reducing power due to radio communications can be achieved by two complementary main objectives: (i) minimizing the output transmit power while maintaining sufficient wireless link quality and (ii) minimizing useless wake-up and channel hearing while still being reactive.
As the physical layer affects all higher layers in the protocol stack, it plays an important role in the energy-constrained design of WSNs. The question to answer can be summarized as: how much signal processing can be added to decrease the transmission energy (i.e. the output power level at the antenna) such that the global energy consumption be decreased? The temporal and spatial diversity of relay and multiple antenna techniques are very attractive due to their simplicity and their performance for wireless transmission over fading channels. Cooperative MIMO (multiple-input and multiple-output) techniques have been first studied in [101] , [109] and have shown their efficiency in terms of energy consumption [98] . Our research aims at finding new energy-efficient cooperative protocols associating distributed MIMO with opportunistic and/or multiple relays and considering wireless channel impairments such as transmitters desynchronisation.
Another way to reduce the energy consumption consists in decreasing the radio activity, controlled by the medium access (MAC) layer protocols. In this regard, low duty-cycle protocols, such as preamble-sampling MAC protocols, are very efficient because they improve the lifetime of the network by reducing the unnecessary energy waste [87] . As the network parameters (data rate, topology, etc.) can vary, we propose new adaptive MAC protocols to avoid overhearing and idle listening.
Finally, MIMO precoding is now recognized as a very interesting technique to enhance the data rate in wireless systems, and is already used in Wi-Max standard (802.16e). This technique can also be used to reduce transmission energy for the same transmission reliability and the same throughput requirement. One of the most efficient precoders is based on the maximization of the minimum Euclidean distance (-) between two received data vectors [94] , but it is difficult to define the closed-form of the optimized precoding matrix for large MIMO system with high-order modulations. Our goal is to derive new generic precoders with simple expressions depending only on the channel angle and the modulation order.