Search

Modern environmental monitoring systems are being envisioned as constellations of tiny smart sensors, densely deployed on the territory to be monitored. Data provided by such sensors is meaningless if it is not equipped with information about the context to which it refers. Such a context is mainly related to the location of sensors, which allows to manage collected data by means of Geographic Information Systems. Due to scarce resources of tiny sensors characteristics, localization services are usually range-free and based on simple measurements provided by the majority of radio chips, such as the Received Signal Strength Indicator (RSSI). However, existing solutions suffer of inaccuracy issues, due to the unreliability of the adopted indicators. This article proposes a range-free localization algorithm based on data fusion, which combines the RSSI with the Power of Transmission (PoT), by estimating the RSSI at different PoT levels. We show, through experimentation on real tiny devices, how this simple solution allows reducing the localization error to about 16% with respect to state-of-art algorithms.

Multicore processors are becoming more and more attractive in embedded and safety-critical domains because they allow increasing the performance by ensuring reduced power consumption. However, moving to multicore systems raises novel dependability challenges: the number of cores, concurrency issues, shared resources and interconnections among cores make it hard to develop and validate software deployed on the top of multicore processors. This paper discusses a preliminary fault injection framework, which aims to investigate dependability properties of multicore-based systems. The proposed framework leverages the error reporting architecture provided by modern processors and has been instantiated in the context of the Intel Core i7 processor. Fault injection campaigns have been conducted under the Linux OS to show the benefits of the framework.

The migration from mobile to ubiquitous Internet is at hand, due to the intense growth of short range wireless technologies. Users accessing the Internet through wireless devices are increasing, if compared to “wired” ones, and they expect the same dependability level they already experience on wired networks, that is high quality “always on” wireless networks. But how can we analyze the dependability level of a wireless network? Direct analysis of failures from the field of application is an effective practice to understand the actual dependability behavior of an operational system. However, despite its wide use over the last four decades on a large variety of systems, field data analysis has rarely been applied to wireless networks. Through the experience gained from extensive failure analysis of Bluetooth networks, the article shows how field failure data can play a key role to fill the gap on understanding the dependability behavior of wireless networks.

The increasing success of mobile-enabled embedded devices is stressing the need for software architectures facing mobility-related issues. This paper proposes a simple yet effective mobility management scheme to ease the development of mobile ubiquitous applications. The scheme seamlessly handles handoff events and provides ubiquitous applications with both location-awareness and mobility prediction support. An implementation prototype has been developed on real-world Bluetooth enabled devices. Experimental results are then obtained from the prototype, showing the effectiveness of the proposed scheme.