[ Energy Profiling and Management for Networked Embedded Systems ]

Energy is a scarce resource in embedded, battery-driven communication devices, such as sensor nodes and cell phones. In this research, we try to understand how embedded applications spend energy. We use the Quanto framework to determine the breakdown of energy among hardware components over time and associate energy consumption with high-level activities. This facilitates the debugging of embedded programs in terms of energy leaks and consumption optimization. Based on our findings, we intend to create a framework that enables energy management for operations between nodes and internal scheduling.

[ Localization Techniques for Mobile Sensor Networks ]

Location awareness has become an important feature for many Wireless Sensor Network (WSN) applications. Examples of such applications include target tracking, emergency services, geographic routing, etc. Due to cost and energy constraints, not all nodes have localization hardware (e.g., GPS receivers). Being this way, localization systems for WSNs usually employ a small set of nodes which are aware of their own coordinates -- anchors or reference points -- and cooperatively distribute location data to regular nodes, helping them estimate their own position.

The early localization algorithms primarily concerned static networks. When mobility is considered, these algorithms often fail to obtain a position estimate with reasonable accuracy. Instead of determining the location directly, an alternative solution is to represent an estimate as a probability distribution over the deployment area. We investigate the usage of probabilistic methods on the localization problem in mobile, error-prone, wireless communication networks.

[ An RFID-based Positioning System for Mobile Ad-hoc Networks ]

RFID technology provides an economically feasible means to build a reliable and scalable localization framework which can be used to implement a wide range of civilian and military applications. Embedding computing and communication capabilities into our physical and social environment imposes unique challenges in terms of hardware and software constraints, and demands the understanding of social behavior in ordinary and emergency settings. In this research, we propose, design and build the software component for an envisioned national infrastructure in Japan, which aims at enabling ubiquitous safety-enhancing services by deploying RFID tags across the nation, developing novel positioning mechanisms, and providing key application services.

[ Intrusion Detection Systems for WSNs ]

Assuming that a sensor network is deployed in an open and unprotected environment, data transmitted by sensor nodes are fatally susceptible to security attacks. Although preventive measures can be used to avoid certain types of attacks, there are situations in which the network, nonetheless, becomes compromised. An intrusion detection system (IDS) acquires information related to attack techniques from malicious nodes inside a network and uses it to discern regular nodes from infringing ones. Later, we can use this monitored data to infer behavior-based attacks and develop prevention systems. We proposed an IDS that fits the demands and restrictions of WSNs.

[ Application Development for WSNs ]

The development of wireless sensor networks (WSNs) was originally motivated by military applications such as battlefield surveillance. However, WSNs are now used in many civilian application areas, including environment and habitat monitoring, healthcare, home automation, and traffic control.

During my first year as a member of SensorNet, my research focused on the development of applications using the Crossbow Mica2 motes. This has culminated into my undergraduate thesis entitled "Vehicle Detection and Classification Using Wireless Sensor Networks" (in Portuguese). I propose a distributed system used for discovery and categorization of automobiles. We can infer such output based on the variance of the Earth's magnetic field induced by the ferrous mass of moving vehicles. Sensor nodes are strategically placed next to the traffic lane and the magnetic signature of each vehicle passing nearby is collected and processed before being sent to a central repository. The central repository serves as a base station, where we can visualize the traffic information.