Marco Mobilio

Falls as anomalies? An experimental evaluation using smartphone accelerometer data

Life expectancy keeps growing and, among elderly people, accidental falls occur frequently. A system able to promptly detect falls would help in reducing the injuries that a fall could cause. Such a system should meet the needs of the people to which is designed, so that it is actually used. In particular, the system should be minimally invasive and inexpensive. Thanks to the fact that most of the smartphones embed accelerometers and powerful processing unit, they are good candidates both as data acquisition devices and as platforms to host fall detection systems. For this reason, in the last years several fall detection methods have been experimented on smartphone accelerometer data. Most of them have been tuned with simulated falls because, to date, datasets of real-world falls are not available. This article evaluates the effectiveness of methods that detect falls as anomalies. To this end, we compared traditional approaches with anomaly detectors. In particular, we experienced the kNN and the SVM methods using both the one-class and two-classes configurations. The comparison involved three different collections of accelerometer data, and four different data representations. Empirical results demonstrated that, in most of the cases, falls are not required to design an effective fall detector.

A Layered Architecture Based on Previsional Mechanisms

The paper presents a layered architecture that improves software modularity and reduces computational and communication overhead for systems requiring data from sensors in order to perform domain-related elaborations (e.g., tracking and surveillance systems). Each layer manages hypotheses that are abductions related to objects modeling the ”real world” at a specific abstraction level, from raw data up to domain concepts. Each layer, by analyzing hypotheses coming from the lower layer, abduces new hypotheses regarding objects at a higher level of abstraction (e.g., from image blobs to identified people) and formulates timed previsions about objects. The failure of a prevision causes a hypothesis to flow up-stream. In turn, previsions can flow downstream, so that their verification is delegated to the lower layers. The proposed architectural patterns have been reified in a Java framework, which is being exploited in an experimental multi-camera tracking system.

SPACES: Subjective sPaces Architecture for Contextualizing hEterogeneous Sources

The growing use of sensors in smart environments applications like smart homes, hospitals, public transportation, emergency services, education, and workplaces not only generates constantly increasing of sensor data, but also rises the complexity of integration of heterogeneous data and hardware devices. Existing infrastructures should be reused under different application domain requirements, applications should be able to manage data coming from different devices without knowing the intrinsic characteristics of the sensing devices, and, finally, the introduction of new devices should be completely transparent to the existing applications. The paper proposes a set of architectural abstractions aimed at representing sensors’ measurements that are independent from the sensors’ technology. Such a set can reduce the effort for data fusion and interpretation, moreover it enforces both the reuse of existing infrastructure and the openness of the sensing layer by providing a common framework for representing sensors’ readings. The abstractions rely on the concepts of space. Data is localized both in a positing and in a measurement space that are subjective with respect to the entity that is observing the data. Mapping functions allow data to be mapped into different spaces so that different entities relying on different spaces can reason on data.

An architecture for the design of platforms supporting responsive environments

Responsive environments are able to sense the environment and to respond to it and to the users that inhabit it. Those systems require both the integration of heterogeneous devices and an abstract representation of the environment to reason about interesting changes. The paper presents DEA (Domain Entities Access), an architecture that enables the realization of platforms supporting responsive environments in the interaction with instrumented physical environments through the observation and the control of meaningful domain entities, thus abstracting from any technological details. Platforms can be easily realized by plugging specific domain-dependant components in a framework that manages all the domain-independent aspects. Thus, the architecture results to be open with respect to both new devices and new typologies of domain entities. A prototypical implementation of the framework has been provided. Moreover, a specific platform has been realized to support an end-user application dealing with instrumented environments.

Time in the Domain Entities Access Architecture

Domain Entities Access is an architecture that enables the realization of platforms supporting responsive environments in the interaction with instrumented physical environments through the observation and the control of meaningful domain entities. This results in an environment model that abstracts from any technological details. Domain entities are characterized by a set of pairs property-value. The value of a property is the last inferred one without any information with respect to when the data used in the inference have been acquired. Thus, the status of domain entities lacks of timeliness. The architecture has been revised so that end-user applications can rely on both inspection and control mechanisms whose results are driven by time. The new implementation of the framework have been validated in a real simplified scenario.