Abstract
While Cluster-Tree network topologies look promising for WSN applications with timeliness and energy-efficiency requirements, we are yet to witness its adoption in commercial and academic solutions. One of the arguments that hinder the use of these topologies concerns the lack of flexibility in adapting to changes in the network, such as in traffic flows. This paper presents a solution to enable these networks with the ability to self-adapt their clusters' duty-cycle and scheduling, to provide increased quality of service to multiple traffic flows. Importantly, our approach enables a network to change its cluster scheduling without requiring long inaccessibility times or the re-association of the nodes. We show how to apply our methodology to the case of IEEE 802.15.4/ZigBee cluster-tree WSNs without significant changes to the protocol. Finally, we analyze and demonstrate the validity of our methodology through a comprehensive simulation and experimental validation using commercially available technology on a Structural Health Monitoring application scenario.

Abstract
Several concurrent applications running on a sensor network may cause a node to transmit packets at distinct periods, which increases the radio-switching rate and has significant impact in terms of the overall energy consumption. We propose to batch the transmissions together by defining a harmonizing period to align the transmissions from multiple applications at periodic boundaries. This harmonizing period is then leveraged to design a distributed protocol called Network-Harmonized Scheduling (NHS) that coordinates transmissions across nodes and provides real-time guarantees in a multi-hop network. Categories and Subject Descriptors-C. 3 [Computer Systems Organization]: Special-purpose and Application-Based Systems: Real-time and embedded systems

Abstract
Smartphones are everywhere, and they are a very attractive platform to perform unobtrusive monitoring of users. In this work, we use common features of modern smartphones to build a human activity recognition (HAR) system for elderly care. We have built a classifier that detects the transport mode of the user including whether an individual is inactive, walking, in bus, in car, in train or in metro. We evaluated our approach using over 24 hours of transportation data from a group of 15 individuals. Our tests show that our classifier can detect the transportation mode with over 90% accuracy.