Christophe Loyez

Passive Acoustic Monitoring (PAM) is widely used for marine mammal observation, but conventional monitoring systems typically require continuous high-power signal acquisition and processing, limiting long-term autonomous deployment. In this work, we present an Always-On Sperm-Whale Detector (AO-SWD) based on a neuromorphic and ultra-low-power processing pipeline combining a bio-inspired artificial cochlea with FPGA-based spiking neural network (SNN) inference. The proposed system uses a 16-channel ultra-low-power (ULP) cochlea implemented in 65 nm CMOS technology operating in the subthreshold regime. The cochlea converts underwater acoustic signals directly into asynchronous spike trains while consuming only 4 μW total power (250 nW/channel). The generated spike-based representation is processed using lightweight feedforward SNN architectures trained with snnTorch and optimized for FPGA deployment using the ModNEF framework. Several network architectures and temporal resolutions were evaluated to study the trade-off between detection accuracy, latency, and hardware complexity. The best software configuration, using a network architecture of 16 input channels, 128 hidden neurons, and 1 output neuron with 20 time steps, achieved 93.30% classification accuracy for sperm whale click detection. The corresponding FPGA implementation achieved 91.15% accuracy after hardware-aware optimization while maintaining real-time operation. The proposed architecture demonstrates the feasibility of combining ultra-low-power neuromorphic sensing with embedded spiking inference for energy-efficient underwater acoustic monitoring. These results highlight the potential of event-driven bio-inspired systems for long-duration autonomous ecological monitoring applications.

Energy efficiency is one of the main concerns in the design of embedded circuits, especially considering the ever-growing amount of portable devices produced for specialized to everyday life applications. Taking inspiration from neuronal processes in the brain, neuromorphic systems are seen as promising solutions to this concern. Great advances in all fields led to the production of numerous hardware implementations, digital or mixed-signal for the most part. While digital systems showcase high accuracy performances and an advanced technological maturity, they fail to reach the ultra-low power (ULP) consumptions of emerging technologies or fully analog implementations due to generally non-dedicated chips and bulky hardware. In this work, we designed and implemented a bioinspired analog demonstrator of inter-pulse delay detection on standard complementary metal oxide semiconductor in the subthreshold operation mode. Relying on the temporal pattern recognition mechanism in female field crickets, our circuit reach on average 750 pW of total power consumption under probes during detection on real-world recordings of male crickets calling song. The circuit was evaluated in quiet, noisy, and multi-source environments, demonstrating strong detection performances given its sparse architecture and ULP consumption.

The development of positioning systems has been significantly advanced by a combination of technological innovations, such as improved sensors, signal processing, and computational power, alongside inspiration drawn from biological mechanisms. Although vision is the main means for positioning oneself—or elements relative to oneself—in the environment, other sensory mediums provide additional information, and may even take over when visibility is lacking, such as in the dark or in troubled waters. In particular, the auditory system in mammals greatly contributes to determining the location of sound sources, as well as navigating or identifying objects’ texture and shape, when combined with echolocation behavior. Taking further inspiration from the neuronal processing in the brain, neuromorphic computing has been studied in the context of sound source localization and echolocation-based navigation, which aim at better understanding biological processes or reaching state-of-the-art performances in energy efficiency through the use of spike encoding. This paper sets out a review of these neuromorphic sound source localization, sonar- and radar-based navigation systems, from their earliest appearance to the latest published works. Current trends and possible future directions within this scope are discussed.

Eight decades ago, the medial superior olive responsible for the extraction of the interaural time difference (ITD) was described as a combination of coincidence detection neurons and delay lines by Jeffress, a model largely used in spiking sound source localization systems. Coincidence detection mechanisms are also found in the visual system to encode speed of motion and time delays as described by the Hassenstein-Reichardt Detector (HRD), a simple model of motion detection derived from behavioral observations in beetles. In this paper, we propose an HRD-based model for ITD extraction of a binaural microphone pair from raw waveforms using neuromorphic components. This model is simulated in computer language Python using equations correctly reflecting an analog neuromorphic technology. With a multilateration method, we estimate the direction of arrival (DOA) of click-like sounds recorded indoor with a 3-microphone rectangular array at distances ranged between 24cm and 10m. A novel simplification of the mathematical equation is introduced for a rectangular array reducing by one order the polynomial characteristic expression giving the sound source’s 2D coordinates. We report an encouraging DOA estimation accuracy of at least 84% (±5°) for sound sources located between 1m and 3m away. The reported work expresses the potential for an ultra-low-power consumption analog implementation of the proposed coincidence detector.

This paper presents a pulse radio transmission system operating in the millimeter-wave regime for simultaneous binary information transmission and location in intra-building environment. In particular, the location solution mutualizes the hardware architectures of three sensors of known position to extract the location information of a sensor entering their vicinity and thus uses the radio communication signals to extract this location information. Experimental validation considering 60GHz transceiver designed and realized on pseudomorphic high electron mobility transistor (pHEMT) technology is exemplary shown with localization error quantified to 16 cm.

Non-contact detection and estimation of vital signs such as respiratory and cardiac frequencies is a powerful tool for surveillance applications. In particular, the continuous wave bio-radar has been widely investigated to determine the physiological parameters in a non-contact manner. Since the RF-reflected signal from the human body is corrupted by noise and random body movements, traditional Fourier analysis fails to detect the heart and breathing frequencies. In this effort, cyclostationary analysis has been used to improve the radar performance for non-invasive measurement of respiratory rate and heart rate. However, the preliminary works focus only on one frequency and do not include the impact of attenuation and random movement of the body in the analysis. Hence in this paper, we evaluate the impact of distance and noise on the cyclic features of the reflected signal. Furthermore, we explore the assessment of second order cyclostationary signal processing performance by developing the cyclic mean, the conjugate cyclic autocorrelation and the cyclic cumulant. In addition, the analysis is carried out using a reduced number of samples to reduce the response time. Implementation of the cyclostationary technique using a bi-static radar configuration at 2.5 GHz is shown as an example to demonstrate the proposed approach.

This work deals with six-port technology applied to a localization system operating in the millimeter-wave frequency band. The proposed system is mainly based on a MMIC 60 GHz six-port circuit which is designed and manufactured with passive mixers, 90° phase-shifters, active and passive dividers to build a sensitive vector detector enabling to determine both Angle Of Arrival (AOA) and Time Difference of Arrival (TDOA). Based on this six-port technology, an original localization technique is implemented to minimize the degradation of the location error.