We propose a distributed recognition framework to classify continuous human actions using a low-bandwidth wearable motion sensor network, called distributed sparsity classifier (DSC). The algorithm classifies human actions using a set of training motion sequences as prior examples. It is also capable of rejecting outlying actions that are not in the training categories. The classification is operated in a distributed fashion on individual sensor nodes and a base station computer. We model the distribution of multiple action classes as a mixture subspace model, one subspace for each action class. Given a new test sample, we seek the sparsest linear representation of the sample w.r.t. all training examples. We show that the dominant coefficients in the representation only correspond to the action class of the test sample, and hence its membership is encoded in the sparse representation. Fast linear solvers are provided to compute such representation via ℓ 1-minimization. To validate the accuracy of the framework, a public wearable action recognition database is constructed, called wearable action recognition database (WARD). The database is comprised of 20 human subjects in 13 action categories. Using up to five motion sensors in the WARD database, DSC achieves state-of-the-art performance. We further show that the recognition precision only decreases gracefully using smaller subsets of active sensors. It validates the robustness of the distributed recognition framework on an unreliable wireless network. It also demonstrates the ability of DSC to conserve sensor energy for communication while preserve accurate global classification.

Event detection is a major issue for applications of wireless sensor networks. In order to detect an event, a sensor network has to identify which application-specific incident has occurred based on the raw data gathered by individual sensor nodes. In this context, an event may be anything from a malfunction of monitored machinery to an intrusion into a restricted area. The goal is to provide high-accuracy event detection at minimal energy cost in order to maximize network lifetime. In this paper, we present a system for collaborative event detection directly on the sensor nodes. The system does not require a base station for centralized coordination or processing, and is fully trainable to recognize different classes of application-specific events. Communication overhead is reduced to a minimum by processing raw data directly on the sensor nodes and only reporting which events have been detected. The detection accuracy is evaluated using a 100node sensor network deployed as a wireless alarm system on the fence of a real-world construction site.

In many problems in computer vision, data in multiple classes lie in multiple low-dimensional subspaces of a highdimensional ambient space. However, most of the existing classification methods do not explicitly take this structure into account. In this paper, we consider the problem of classification in the multi-subspace setting using sparse representation techniques. We exploit the fact that the dictionary of all the training data has a block structure where the training data in each class form few blocks of the dictionary. We cast the classification as a structured sparse recovery problem where our goal is to find a representation of a test example that uses the minimum number of blocks from the dictionary. We formulate this problem using two different classes of non-convex optimization programs. We propose convex relaxations for these two non-convex programs and study conditions under which the relaxations are equivalent to the original problems. In addition, we show that the proposed optimization programs can be modified properly to also deal with corrupted data. To evaluate the proposed algorithms, we consider the problem of automatic face recognition. We show that casting the face recognition problem as a structured sparse recovery problem can improve the results of the state-of-the-art face recognition algorithms, especially when we have relatively small number of training data for each class. In particular, we show that the new class of convex programs can improve the state-ofthe-art face recognition results by 10 % with only 25 % of the training data. In addition, we show that the algorithms are robust to occlusion, corruption, and disguise. 1.

Abstract—Activities of daily living are important for assessing changes in physical and behavioural profiles of the general population over time, particularly for the elderly and patients with chronic diseases. Although accelerometers are widely integrated with wearable sensors for activity classification, the positioning of the sensors and the selection of relevant features for different activity groups still pose interesting research challenges. This paper investigates wearable sensor placement at different body positions and aims to provide a framework that can answer the following questions: (i) What is the ideal sensor location for a given group of activities? (ii) Of the different time-frequency features that can be extracted from wearable accelerometers, which ones are most relevant for discriminating different activity types? Index Terms—Wearable sensors, Body sensor networks, feature selection, sensor positioning.