Probability Distributions.- Linear Models for Regression.- Linear Models for Classification.- Neural Networks.- Kernel Methods.- Sparse Kernel Machines.- Graphical Models.- Mixture Models and EM.- Approximate Inference.- Sampling Methods.- Continuous Latent Variables.- Sequential Data.- Combining Models.

Pattern Recognition and Machine Learning

Data Mining Practical Machine Learning Tools and Techniques

This paper presents and evaluates two principles for wireless routing protocols. The first is datapath validation: data traffic quickly discovers and fixes routing inconsistencies. The second is adaptive beaconing: extending the Trickle algorithm to routing control traffic reduces route repair latency and sends fewer beacons.We evaluate datapath validation and adaptive beaconing in CTP Noe, a sensor network tree collection protocol. We use 12 different testbeds ranging in size from 20--310 nodes, comprising seven platforms, and six different link layers, on both interference-free and interference-prone channels. In all cases, CTP Noe delivers > 90% of packets. Many experiments achieve 99.9%. Compared to standard beaconing, CTP Noe sends 73% fewer beacons while reducing topology repair latency by 99.8%. Finally, when using low-power link layers, CTP Noe has duty cycles of 3% while supporting aggregate loads of 30 packets/minute.

/pdf/collection-tree-protocol-selhhlblmq.pdf

Collection tree protocol

Advances in wireless communication technologies, such as wearable and implantable biosensors, along with recent developments in the embedded computing area are enabling the design, development, and implementation of body area networks. This class of networks is paving the way for the deployment of innovative healthcare monitoring applications. In the past few years, much of the research in the area of body area networks has focused on issues related to wireless sensor designs, sensor miniaturization, low-power sensor circuitry, signal processing, and communications protocols. In this paper, we present an overview of body area networks, and a discussion of BAN communications types and their related issues. We provide a detailed investigation of sensor devices, physical layer, data link layer, and radio technology aspects of BAN research. We also present a taxonomy of BAN projects that have been introduced/proposed to date. Finally, we highlight some of the design challenges and open issues that still need to be addressed to make BANs truly ubiquitous for a wide range of applications.

Body Area Networks: A Survey

Multi-sensor fusion in body sensor networks: State-of-the-art and research challenges

There is a strong medical need for continuous, unobstrusive respiratory monitoring, and many shortcomings to existing methods. Previous work shows that MEMS accelerometers worn on the torso can measure inclination changes due to breathing, from which a respiratory rate can be obtained. There has been limited validation of these methods. The problem of practical continuous monitoring, in which patient movement disrupts the measurements and the axis of interest changes, has also not been addressed. We demonstrate a method based on tri-axial accelerometer data from a wireless sensor device, which tracks the axis of rotation and obtains angular rates of breathing motion. The resulting rates are validated against gyroscope measurements and show high correlation to flow rate measurements using a nasal cannula. We use a movement detection method to classify periods in which the patient is static and breathing signals can be observed accurately. Within these periods we obtain a close match to cannula measurements, for both the flow rate waveform and derived respiratory rates, over multi-hour datasets obtained from wireless sensor devices on hospital patients. We discuss future directions for improvement and potential methods for estimating absolute airflow rate and tidal volume.

Respiratory Rate and Flow Waveform Estimation from Tri-axial Accelerometer Data

This paper introduces SpeckMAC, a novel low-power distributed, unsynchronised, random-access MAC protocol for a wireless mobile ad-hoc network of miniature specks called specknet. Two variations of the SpeckMAC protocol were compared theoretically with the well-known B-MAC protocol. All three MAC protocols were implemented on a larger speck prototype called the ProSpeckz for a logical location maintenance algorithm for two types of batteries with differing current drain profiles. ProSpeckz utilizing SpeckMAC-B and SpeckMAC-D showed improvements in lifetime over those using the B-MAC protocol by 27.4% and 38.0%, respectively when Polymer Li-ion batteries were used; and by 83.5% and 117.9%, respectively, when CR1220 button cells were employed.

SpeckMAC: low-power decentralised MAC protocols for low data rate transmissions in specknets

A realtime posture tracking system has been developed using a network of compact wireless sensor devices worn by the user. Each device is a complete inertial/magnetic tracking unit which performs in situ orientation estimation based on its own sensor readings, using a complementary quaternion-based filter. Compared to existing systems which transmit raw sensor data to a PC for processing, it is shown that this technique reduces bandwidth requirements by 79% for typical usage. In combination with a time division multiple access scheme, this reduction allows for full-body tracking using 15 devices at a 64Hz update rate through a single 250kbps receiver. The data is applied to a rigid body model of the subject to provide a realtime display, and can be exported for use in major animation packages.

/pdf/orient-2-a-realtime-wireless-posture-tracking-system-using-2pn8cbcp73.pdf

Orient-2: a realtime wireless posture tracking system using local orientation estimation

Speckled Computing is an emerging technology in which data will be sensed in minute (ultimately around one cubic millimetre) semiconductor grains called Specks. Information will be extracted, exchanged and processed in a collaborative fashion in a wireless network of thousands of Specks, called a Specknet. The impact of Speckled Computing on consumer electronics, especially future information appliances, will be disruptive. Objects and the surrounding environment when treated with Specks, or “Speckled”, will be endowed with sensing, processing and wireless networking capabilities. This would effectively “smarten” everyday objects and surroundings post hoc, transforming them into networked information appliances. This paper introduces the concepts of Specks, Specknets and Speckled Computing, and outlines the challenges to be overcome to realise this technology. A prototype for Specks called ProSpeckz (Programmable Specks over Zigbee Radio) which is currently used as a rapid development platform for Speckled Computing is described. ProSpeckz is also intended as an enabler for integrating the technology of Speckled Computing into consumer electronics applications and some illustrative examples are described in this paper. Index Terms — Networked Information Appliances, Smart Home applications and protocols, Speckled Computing, Wireless computational networks

Speckled computing: disruptive technology for networked information appliances

The use of wireless devices with accelerometers and gyroscopes to measure the movements of humans and objects is a growing area of interest. Applications range from simple activity detection to detailed full-body motion capture using networks of sensors worn on the body. A variety of algorithms have been proposed for these applications, but opportunities for accurate evaluation and comparison have been limited due to the many difficulties with performing rigorous experiments. We present a simulation environment, specifically for inertial sensing applications, designed to tackle this problem. We simulate sensor readings based on continuous trajectory models, and show how suitable models can be generated from existing motion capture or other sampled data. We show a good match between our simulated data and real sensor data for human movements. We also model a wide range of real-world issues such as non-ideal sensors, magnetic field distortions, timing factors and radio packet losses. To demonstrate the capabilities of our simulator, we present new results comparing four existing orientation estimation algorithms for human motion capture.

D. K. Arvind

Papers

Respiratory Rate and Flow Waveform Estimation from Tri-axial Accelerometer Data

SpeckMAC: low-power decentralised MAC protocols for low data rate transmissions in specknets

Orient-2: a realtime wireless posture tracking system using local orientation estimation

Speckled computing: disruptive technology for networked information appliances

IMUSim: A simulation environment for inertial sensing algorithm design and evaluation