Homepage Inria website

Section: New Results

Interaction between Algorithms and Architectures

Sensor-Aided Non-Intrusive Load Monitoring

Participants : Xuan-Chien Le, Olivier Sentieys.

Non-Intrusive Load Monitoring (NILM) plays an important role in energy management and energy reduction in buildings and homes. An NILM system does not need a large amount of deployed power meters to monitor the power usage of home devices. Instead, only one meter on the main power line is necessary to detect and identify the operating devices. There are many approaches to solve the problem of device determination in NILM. The features applied in low-frequency based approach essentially include the step-change (or edge) and the steady state. In [47] we introduced three algorithms to solve the l1-norm minimization problem in NILM and results on power measurements obtained from a real appliance deployment. With a small number of devices, the obtained precision varies from 75% to 99%, depending on the tolerance criterion to determine the steady state of a given device.

Posture and Gesture Recognition using Wireless Body Sensor Networks

Participants : Arnaud Carer, Alexis Aulery, Olivier Sentieys.

The BoWi project (Body Wold Interactions) aims at designing a Wireless Body Sensor Network (WBSN) for accurate Gesture and Body Movement estimation with extremely severe constraints in terms of footprint and power consumption. Advantages of such system mainly come from its possible use in indoor or outdoor environments without any additional equipment. The 3D geolocation approach will combine radio communication distance measurement and inertial sensors and it will also strongly benefit from cooperative techniques based on multiple observations and distributed computation. Different types of applications, as health care, activity monitoring and environment control, are considered and evaluated along with a human-machine interface expertise.

In [32] we presented three different use cases of WBSN for posture and gesture recognition developed by increasing demands in terms of accuracy: posture recognition, gesture recognition and motion capture. This work is based on a simulator designed to explore algorithmic solutions for posture and gesture identification. Simulation results were performed with a set of different algorithm and sensor proposals for three usages including a Principal Component Analysis (PCA) for posture classification. We show how sensor and algorithm can be carefully chosen according to application scenarios while minimising implementation complexity.

For applications based on predefined postures such as environment control and physical rehabilitation, we show in [31] that low cost and fully distributed solutions, that minimize radio communications, can be efficiently implemented. Considering that radio links provide distance information, we also demonstrate that the matrix of estimated inter-node distances offers complementary information that allows for the reduction of communication load. Our results are based on a simulator that can handle various measured input data, different algorithms and various noise models. Simulation results are useful and used for the development of real-life prototype.

Energy Harvesting and Power Management

Participants : 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.

In [24] , 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 system lifetime but also to enable robust booting from the exhausting energy of the system. Simulations and experiments show that our multiple-energy-sources converter (MESC) can achieve booting time in order of seconds. Although capacitors have virtual recharge cycles, they suffer 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 those storage devices satisfying energy-neutral operation (ENO) and maximizing system quality-of-service (QoS) in EH nodes, when using a given energy source, was also proposed. Experiments validating this framework are performed on a real WSN platform with both photovoltaic cells and thermal generators in an indoor environment. Moreover, simulations on OMNET++ showed that the energy storage optimized from our design framework is used up to 93.86%.

A Power Manager (PM) is usually embedded in EH wireless nodes to adapt the computation load by changing their wake-up interval according to the harvested energy. In order to prolong the network lifetime, the PM must ensure that every node satisfies the Energy Neutral Operation (ENO) condition. However, when a multi-hop network is considered, changing the wake-up interval regularly may cripple the synchronization among nodes and therefore, degrade the global system Quality of Service (QoS). In [25] , a Wake-up Variation Reduction Power Manager (WVR-PM) was proposed to solve this issue. This PM is applied for wireless nodes powered by a periodic energy source (e.g. light energy in an office) over a constant cycle of 24 hours. Not only following the ENO condition, our power manager also reduces the wake-up interval variations of WSN nodes. Based on this PM, an energy-efficient protocol, named Synchronized Wake-up Interval MAC (SyWiM), was also proposed. OMNET++ simulation results using three different harvested profiles show that the data rate of a WSN node can be increased up to 65% and the latency reduced down to 57% compared to state-of-the-art PMs. Validations on a real WSN platform have also been performed and confirmed the efficiency of our approach.

Signal Processing for High-Rate Optical Communications

Participants : Trung-Hien Nguyen, Olivier Sentieys, Arnaud Carer.

Mary quadrature amplitude modulation (m-QAM) combined with coherent detection and digital signal processing (DSP) is a promising candidate for the implementation of next generation optical transmission systems. However, as the number of modulation levels increases, the sensitivity to system imperfections such as phase noise of the transmitter and the local oscillator lasers or fiber nonlinearities is exacerbated. Moreover, the amplitude and phase imbalances between the in-phase (I) and quadrature (Q) channels in the transmitter (Tx) and the front-end of the receiver (Rx), which is often referred to as IQ imbalance, is also troublesome if not compensated

In [52] , we proposed a novel simple blind adaptive IQ imbalance compensation based on a decision-directed least-mean-square (DD-LMS) algorithm integrated to a modified butterfly FIR filter configuration. Since only 2 FIR filter coefficients-sets are used, instead of 4 in the conventional configuration, the time for updating the coefficients and the hardware resources (such as coefficient memories and number of look-up tables) in real time field-programmable gate array (FPGA) platforms is then reduced using this method. A reduction in hardware complexity by a factor of about 3 is achieved by the proposed joint method. The proposed structure is experimentally validated with a 40Gbit/s 16-QAM signal. A 7‑dB power penalty reduction is experimentally achieved at a bit error rate (BER) of 10-3 in the presence of a 10 degree phase imbalance, confirming the effectiveness of the proposed algorithm. The equalization capability remains even in the presence of group velocity dispersion along the link, which is numerically confirmed with optical fiber transmission up to 1200 km and 20 phase imbalance.

In [50] , circular harmonic expansion-based carrier frequency offset estimation was investigated for optical m-QAM communication systems. The proposed method, combined with a gradient-descent algorithm, shows better performance compared to already proposed VVMFOE and 4th-power methods.