This paper serves as a survey and empirical evaluation of the state-of-the-art in activity recognition methods using accelerometers. The paper is particularly focused on long-term activity recognition in real-world settings. In these environments, data collection is not a trivial matter; thus, there are performance trade-offs between prediction accuracy, which is not the sole system objective, and keeping the maintenance overhead at minimum levels. We examine research that has focused on the selection of activities, the features that are extracted from the accelerometer data, the segmentation of the time-series data, the locations of accelerometers, the selection and configuration trade-offs, the test/retest reliability, and the generalisation performance. Furthermore, we study these questions from an experimental platform and show, somewhat surprisingly, that many disparate experimental configurations yield comparable predictive performance on testing data. Our understanding of these results is that the experimental setup directly and indirectly defines a pathway for context to be delivered to the classifier, and that, in some settings, certain configurations are more optimal than alternatives. We conclude by identifying how the main results of this work can be used in practice, specifically in experimental configurations in challenging experimental conditions.

Synchronized communication has recently emerged as a prime option for low-power critical applications. Solutions such as Glossy or Time Slotted Channel Hopping (TSCH) have demonstrated end-to-end reliability upwards of 99.99%. In this context, the IETF Working Group 6TiSCH is currently standardizing the mechanisms to use TSCH in low-power IPv6 scenarios. This paper identifies a number of challenges when it comes to implementing the 6TiSCH stack. It shows how these challenges can be addressed with practical solutions for locking, queuing, scheduling and other aspects. With this implementation as an enabler, we present an experimental validation and comparison with state-of-the-art MAC protocols. We conduct fine-grained energy profiling, showing the impact of link-layer security on packet transmission. We evaluate distributed time synchronization in a 340-node testbed, and demonstrate that tight synchronization (hundreds of microseconds) can be achieved at very low cost (0.3% duty cycle, 0.008% channel utilization). We finally compare TSCH against traditional MAC layers: low-power listening (LPL) and CSMA, in terms of reliability, latency and energy. We show that with proper scheduling, TSCH achieves by far the highest reliability, and outperforms LPL in both energy and latency.

Smart health home systems and assisted living architectures rely on severely energy-constrained sensing devices, such as wearable sensors, for the generation of data and their reliable wireless communication to a central location. However, the need for recharging the battery regularly constitutes a maintenance burden that hinders the long-term cost-effectiveness of these systems, especially for health-oriented applications that target people in need, such as the elderly or the chronically ill. These sensing systems generate raw data that is processed into knowledge by reasoning and machine learning algorithms. This paper investigates the benefits of embedded machine learning, i.e. executing this knowledge extraction on the wearable sensor, instead of communicating abundant raw data over the low power network. Focusing on a simple classification task and using an accelerometer-based wearable sensor, we demonstrate that embedded machine learning has the potential to reduce the radio and processor duty cycle by several orders of magnitude; and, thus, substantially extend the battery lifetime of resource-constrained wearable sensors.

Time Slotted Channel Hopping (TSCH) is a link layer protocol defined in the IEEE 802.15.4 standard. Although it is designed to provide highly reliable and efficient service targeting industrial automation systems, scheduling TSCH transmissions in the time and frequency dimensions is left to the implementers. We evaluate the performance of existing autonomous scheduling approaches for TSCH on various traffic patterns and network configurations. We thoroughly investigate the pros and cons of each scheme; moreover, we propose the use of node based channel allocation to improve the performance of the best scheme, and demonstrate its practicality and reliability, with up to 6 percentage points better packet delivery ratio than the second best option while retaining a similar radio duty cycle. Finally, based on our extensive performance evaluation, we provide some guidelines on how to select a scheduler for a given network.