Section: New Results

Interaction between Algorithms and Architectures

Cooperative-cum-Constrained Maximum Likelihood Algorithm for UWB-based Localization in Wireless BANs

Participants : Antoine Courtay, Matthieu Gautier, Gia Minh Hoang [Master's Student] .

Wireless Body Area Network (BAN) is a mainstream technology for numerous application fields (medicine, security, sport science, etc.) and precise determination of wireless sensors' positions responses to the great needs in many applications. This study leverages Ultra Wide Band (UWB) radio which is an attractive technology to achieve the centimeter-level distance measurements. However, the aggregation of the distance information remains a challenge to achieve an accurate localization in wireless BAN. To this aim, we have proposed a novel Cooperative-cum-Constrained Maximum Likelihood (CCML) localization algorithm. This algorithmic study shows the improvement that could be achieved by combining UWB radio and dedicated algorithms. Future works is to integrate UWB technology in the second version of the Zyggie platform developed in Cairn .

MIMO Systems and Cooperative Strategies for Low-Energy Wireless Networks

Participants : Olivier Berder, Olivier Sentieys, Baptiste Vrigneau, Viet-Hoa Nguyen.

Since a couple of years, the Cairn team has reached a significant expertise in multi-antenna systems, especially in linear precoding. If this technique is traditionally used in a collocated way, it could also be used for wireless sensor networks (WSN) in a distributed manner. We presented a new approach, named distributed max-dmin precoding (DMP). This protocol is based on the deployment of a virtual 2x2 max-dmin precoding over one source, one forwarding relay, both equipped with one antenna and a destination involving two antennas. In this context, two kinds of relaying, amplify and forward or decode and forward protocols, were investigated. The performance evaluation in terms of Bit-Error-Rate (BER) and energy efficiency was compared with non cooperative techniques (SISO, SIMO) and the distributed space time block code (STBC) scheme. Our investigations showed that the DMP takes the advantage in terms of energy efficiency from medium transmission distances.

A receiver initiated cooperative medium access control (RIC-MAC) protocol was also proposed for cooperative communications to reduce the energy consumption of WSN. Considering a real WSN platform, the simulation results show that using the proposed RIC-MAC protocol in cooperative communications provides latency and energy gains as compared to multi-hop communications. Even if the energy gain is shown to be reduced when the network traffic load increases, our protocol still brings an energy gain about 22% at 1 packet/second. Finally, considering the impact of traffic load on energy consumption and latency, RIC-MAC is illustrated to be robust to traffic load variations in terms of latency [66] .

Adaptive protocols for Wireless Sensor Networks

Participants : Olivier Berder, Matthieu Gautier, Nhat-Quang Nhan [Master's Student] , Van-Thiep Nguyen.

As tiny sensor nodes are equipped with limited battery, the optimization of the power consumption of these devices is extremely vital. In typical WSN platforms, the radio transceiver consumes major proportion of the energy. Major concerns are therefore to decrease the radio activity by designing efficient MAC protocols.

Energy consumption plays an important role in the design of Wireless Body Area Sensor Network (WBASN). Unfortunately, the performance of WBASNs decreases in high interference environments such as the Industrial, Scientific and Medical (ISM) band where wireless spectrums are getting crowded. In this study [59] , an energy-efficient Medium Access Control (MAC) protocol named C-RICER (Cognitive-Receiver Initiated CyclEd Receiver) is specifically designed for WBASN to cognitively work in high interference environment. C-RICER protocol adapts both transmission power and channel frequency to reduce the interferences and thus, the energy consumption. The protocol is simulated with the OMNET++ simulator. Simulation results show that, depending on the interference level, C-RICER is able to outperform the traditional RICER protocol in terms of energy consumption, packet delay, and network throughput.

In recent years, many MAC protocols for Wireless Sensor Networks (WSNs) have been proposed and evaluated using Matlab simulator and/or network simulators (OMNeT++, NS2, etc.). However, most of them have a static behavior and few network simulations are available for adaptive protocols. Specially, in OMNeT++/MiXiM, there is few energy-efficient MAC protocol for WSNs (B-MAC and L-MAC) and no adaptive protocol. To this end, the TAD-MAC (Traffic Aware Dynamic MAC) protocol has been simulated in OMNeT++ with the MiXiM framework [57] . The simulation results have been used to compare with B-MAC and L-MAC protocol, showing the gain brought by TAD-MAC.

Energy Harvesting and Power Management

Participants : Olivier Berder, Olivier Sentieys, Arnaud Carer, Trong-Nhan Le.

To design autonomous Wireless Sensor Networks (WSNs) with a theoretical infinite lifetime, energy harvesting (EH) techniques have been recently considered as promising approaches. Ambient sources can provide everlasting additional energy for WSN nodes and exclude their dependence on battery. An efficient energy harvesting system which is compatible with various environmental sources such as light, heat or wind energy was proposed. Our platform takes advantage of double-level capacitors not only to prolong the system lifetime but also to enable robust booting from the exhausting energy of the system. Simulations and experiments showed that it can achieve booting time in order of seconds. Although capacitors have virtual recharge cycles, they suffer from higher leakage compared to rechargeable batteries. Increasing their size can decrease the system performance due to leakage energy. Therefore, an energy neutral design framework providing a methodology to determine the minimum size of the storage devices satisfying Energy Neutral Operation (ENO) and maximizing system Quality of Service (QoS) in EH nodes when using a given energy source was proposed. Experiments validating this framework were performed on a real WSN platform with both photovoltaic cells and thermal generators in an indoor environment [30] .

A new PM for EH-WSNs scavenging energy from periodic sources, i.e., ambient energy is not available during the full harvesting cycle, was proposed. Not only respecting the ENO condition, our PM is able to balance the Quality of Service (QoS) during the whole cycle to provide regular data tracking, which is essential for WSN applications like monitoring. Simulations on OMNET++ show that our PM can improve the QoS during the absence of energy by a factor up to 84% compared to state-of-the-art PMs, while guaranteeing the same global QoS [54] .

Multimedia Processing

Participant : Pascal Scalart.

Most noise reduction methods for multimedia signals are usually based on the application of a short-time Wiener filter (MMSE) that is generally expressed as a spectral gain depending on the local signal-to-noise ratio (SNR) on each frequency bin. To estimate such filter, several algorithms can be found in the literature but these conventional approaches lead to a biased estimator for the a priori signal-to-noise estimate. To reduce this bias, we have proposed in [26] a new strategy that relies on the introduction of a correction term in the computation of the Wiener filter depending on the current state of both the available a priori and a posteriori SNR estimates. The proposed solution leads to a bias-compensated a priori SNR estimate, and allows to finely estimating the target signal that is very close to the original noise-free reference. Such refinement procedure has been tested under various noisy environments and show the superiority of the proposed strategy compared to competitive algorithms.

Audio classification systems have recently gained interest for the design of various real-world multimedia services such as audio database indexing with musical genre classification, video indexing using the soundtrack or context awareness. A large majority of audio classification systems can be viewed as offline applications in the sense that there is no strong restriction about how the signal to be classified is accessed. In [44] , we investigate the case where the classification task is performed in real-time in a low-latency classification framework. We proposed different methodologies for the use of feature integration that are based on three key aspects: the selection of the features which have to be temporally integrated, the choice of the integration techniques, i.e. how the temporal information is extracted, and the size of the integration window. The experiments carried out for the classification task show that these different methodologies have a significant impact on the global performance even with the low-latency constraints. In addition, we investigate the detection of howlings that arise in audio signals in [43] . To do so, the processing algorithm is based on a Support Vector Machine (SVM) model in the decision stage and on the combination of energy-based features and also a new feature related to the frequency stability of a howling component. The proposed method can be used in different situation since its provides good results with a very low false alarm rate for a wide range of experimental conditions.

Non-Intrusive Load Monitoring

Participants : Olivier Sentieys, Baptiste Vrigneau, Xuan Chien Le.

Natural resource preservation has recently become a significant concern and has therefore motivated many research and development efforts for energy consumption management in buildings and homes. Efficiently reducing energy consumption at home, work or in a factory, could be afforded by mixing different technologies to not only reduce the energy consumed by consumers, but also to adapt (manage) the energy consumed to the energy that is produced. SMART 2020 outlined the opportunity to capture savings of both energy and Greenhouse Gas (GHG) emissions in 2020, through a range of actions developed by the Information and Communications Technologies (ICT) sector. Smart Grid, Smart Buildings, and Green ICT have the main impact on energy savings. At the energy production side, the electrical grid infrastructure is comprised of three elements: power generation, transmission, and distribution. Electrical power generation consists mainly of the power plants but also includes more and more renewable sources such as wind power or solar panels on energy farms or locally on top of buildings. The cost of energy storage is very high, and hence the current practice is to match energy consumption closely with energy generation, which is more and more fluctuating: challenges could be seen as being able to use energy when the wind blows or the sun shines, and also to avoid the strong power consumption peaks due to people's life. A typical example at home could be to automatically use the dryer when energy is available and therefore cheap, and is now well defined as Smart Grid technologies. At the energy consumption side, the main objective is of course to reduce energy consumption of the different subsystems. Interior lighting, office equipment, heating, cooling, and ventilation make up of more than 85% of the total electricity use and the reduction effort should therefore be concentrated on these systems. For energy management and reduction in homes or building a key enabler is the use of wireless sensor networks to monitor the environment (temperature, activity of people, power consumption of equipment, light, etc.) and to act on subsystems (decrease room temperature, stop or start an equipment, adjust cooling or ventilation, etc.). This is the emerging field of Smart Building Automation.

The objective of this work is strongly linked to the usage of these WSN nodes in the context of smart monitoring of energy consumption and environment (temperature, activity, light). We will propose new Indirect Power Monitoring techniques which enable to estimate energy consumed in a building or in a home without effectively measuring the power consumed. A typical AC smart meter is costly equipment and we therefore want to propose cheap and non-invasive sensor nodes. As an example, to estimate the power consumed by the TV, it is not necessary to measure precisely the current it consumed, but a simple sensor able to recognize that TV is on or off can do the same job with a far less complexity. Another example is the development and deployment of room occupancy and people activity sensors that can lead to significant reduction of the energy by regulating HVAC (Heating, Ventilation and Air-Conditioning) or by switching lights and office equipment. The wireless transmission is the main reason of consuming energy and the new algorithms will propose to make the sensors to cooperate inside a low-distance cluster (an office for example). The algorithms will decide the best strategy and the best information to send back in order to offer the best trade-off between Performance/Complexity/Consumption. This work is closely links to power management techniques and energy harvesting (in-door light, heat, vibration). A power manager embedded in energy harvesting WSN nodes adapts the power consumption and computation loads according to the harvested energy to obtain a theoretically infinite lifetime. The main advantage of using energy harvesting (EH) in the context of building and home monitoring is to avoid battery replacement and therefore to reduce installation and maintenance costs of the system.