Hardware-Based Hopfield Neuromorphic Computing for Fall Detection.

With the popularity of smart wearable systems, sensor signal processing poses more challenges to machine learning in embedded scenarios. For example, traditional machine-learning methods for data classification, especially in real time, are computationally intensive. The deployment of Artificial Intelligence algorithms on embedded hardware for fast data classification and accurate fall detection poses a huge challenge in achieving power-efficient embedded systems. Therefore, by exploiting the associative memory feature of Hopfield Neural Network, a hardware module has been designed to simulate the Neural Network algorithm which uses sensor data integration and data classification for recognizing the fall. By adopting the Hebbian learning method for training neural networks, weights of human activity features are obtained and implemented/embedded into the hardware design. Here, the neural network weight of fall activity is achieved through data preprocessing, and then the weight is mapped to the amplification factor setting in the hardware. The designs are checked with validation scenarios, and the experiment is completed with a Hopfield neural network in the analog module. Through simulations, the classification accuracy of the fall data reached 88.9% which compares well with some other results achieved by the software-based machine-learning algorithms, which verify the feasibility of our hardware design. The designed system performs the complex signal calculations of the hardware's feedback signal, replacing the software-based method. A straightforward circuit design is used to meet the weight setting from the Hopfield neural network, which is maximizing the reusability and flexibility of the circuit design.

Compressive sensing scalp EEG signals: implementations and practical performance.

Highly miniaturised, wearable computing and communication systems allow unobtrusive, convenient and long term monitoring of a range of physiological parameters. For long term operation from the physically smallest batteries, the average power consumption of a wearable device must be very low. It is well known that the overall power consumption of these devices can be reduced by the inclusion of low power consumption, real-time compression of the raw physiological data in the wearable device itself. Compressive sensing is a new paradigm for providing data compression: it has shown significant promise in fields such as MRI; and is potentially suitable for use in wearable computing systems as the compression process required in the wearable device has a low computational complexity. However, the practical performance very much depends on the characteristics of the signal being sensed. As such the utility of the technique cannot be extrapolated from one application to another. Long term electroencephalography (EEG) is a fundamental tool for the investigation of neurological disorders and is increasingly used in many non-medical applications, such as brain-computer interfaces. This article investigates in detail the practical performance of different implementations of the compressive sensing theory when applied to scalp EEG signals.

Compressive sensing: from "compressing while sampling" to "compressing and securing while sampling".

In a traditional signal processing system sampling is carried out at a frequency which is at least twice the highest frequency component found in the signal. This is in order to guarantee that complete signal recovery is later on possible. The sampled signal can subsequently be subjected to further processing leading to, for example, encryption and compression. This processing can be computationally intensive and, in the case of battery operated systems, unpractically power hungry. Compressive sensing has recently emerged as a new signal sampling paradigm gaining huge attention from the research community. According to this theory it can potentially be possible to sample certain signals at a lower than Nyquist rate without jeopardizing signal recovery. In practical terms this may provide multi-pronged solutions to reduce some systems computational complexity. In this work, information theoretic analysis of real EEG signals is presented that shows the additional benefits of compressive sensing in preserving data privacy. Through this it can then be established generally that compressive sensing not only compresses but also secures while sampling.