Sensing Location in the Pocket

We present an approach for recognizing location transitions of persons in buildings, using inertial sensor data from mobile devices. By normalizing trajectories using principal component analysis (PCA), our approach is robust to changes in sensor placement and orientation. On a data set containing 10 location transitions and 7 different placements/orientations of the mobile device, we achieve classification rates of about 95% in average. Moreover, when used in an online mode, we can predict the target location of the user with 80% certainty after the user has covered 35% of the path distance, on average.

Going My Way: a user-aware route planner

Going My Way is a mobile user-aware route planner. The system collects GPS data of a user's everyday locations and provides directions from an automatically selected set of landmarks that are close to the destination from the user's frequently visited locations. In this extended abstract, we present a brief description of the system, the result of a preliminary experiment in memory and recognition of landmarks, and some results of a user evaluation of the system.

Public Icon Communication Service System: A Human Readable Tag Interface for Context-awareness Service

In this paper, we present a public icon communication system (PICSS). The proposed public icon is based on ISO 7001: Public Information Symbols. The PICSS decodes the public icon, determines the service type, then combines it with the user profile and context, and sends a query requesting a service. The visual expression of the public icon provides intuitive human-readability which overcomes the machine-readable codes. Because of the human-readable icon, foreigners could benefit from PICSS. We implemented PICSS image-based recognition system.

Massive - An Adaptive Shopping Assistant

We present Massive, a prototype of a web-based intelligent shopping assistant for mobile devices.

Ad-hoc Grouping of Sensor Nodes that Move Together

We present a method to identify multiple sensor nodes that move together with a specific node in an ad-hoc manner. An agglomerative clustering (AHC) has been applied every time step to find the group members. A threshold to "cut" a branch of the group is also determined in an unsupervised way. The grouping allows a system to monitor who is/are utilizing or carrying what object(s).

A Labeling Tool for a Sensor Networked Environment with a Real-world Knowledge Construction Model

Construction of real-world knowledge is required for detecting real-world events occurring in a networked sensor environment. Since it is difficult to select suitable events to be recognized in a sensor environment a priori, we propose an extensive model for constructing real-world knowledge that is based on a manually labeled event corpus. The proposed model simultaneously focuses on both extending the ontology of real-world events and improving the sensor system. A labeling tool is developed in accordance with the model to support ontology and corpus construction and is being evaluated in a practical labeling experiment.

From Feedback to Awareness: An Ambient Water Consumption Display

Today, water usage in general is of concern as water is becoming a more and more scarce resource worldwide. The task is to make people more aware of their water consumption. This awareness is necessary to change one's water consumption behavior and consequently be able to save water in the long term. Problems with existing water meters are, on the one hand, the place they are installed in. In general, they are not easily accessible and thus the meter is seldom read off. On the other hand, reading these meters requires a high cognitive load. Therefore, conventional meters cannot deliver appropriate direct feedback. The suggested solution proposes the usage of ambient displays directly in the spatial context of water consumption. The advantage of ambient displays is that they are aesthetically pleasing objects and deliver feedback in an unobtrusive and subtle way. The application developed addresses the daily morning shower. Since this shower can take from less than 5 up to more than 30 minutes, there is great potential of creating awareness and saving water. Through the use of an ambient display, the imaginary water level if the drain was closed can be visualized. This can help to create a perception of how much water is actually consumed.

Activity-based Authentication by Ambient Wi-Fi Fingerprint Sensing

Preserving a good balance between security and usability is often an important issue in many ubiquitous computing applications. This paper proposes a new user interface model for security based on device's activity history by environmental Wi-Fi fingerprint sensing. If a device periodically senses and records Wi-Fi fingerprint, such record represents the device's location and activity history. This information can be used to detect whether this device is in a normal situation or in an unusual (abnormal) situation. If two devices compare these activity logs, it is also possible to determine whether these devices are moved together or not. Then the system changes its security level based on such activity information. Unlike other Wi-Fi positioning systems, geological Wi-Fi access point information is not always necessary, because only fingerprint matching is enough for authentication purpose This feature makes our security model more scalable; it works even though a Wi-Fi access point location database is not provided. (This paper describes new user interface model for security based on this idea, and reports initial experimental results.

Reducing Fuel Consumption by Providing In-situ Feedback on the Impact of Current Driving Actions

Reducing the fuel consumption of cars is seen as a central element in reducing CO2 production by individuals. The automotive industry currently focuses on more efficient engines and alternative power concepts. Our research complements this by an assessment of how drivers can reduce fuel consumption of their vehicle by changing their behavior. An initial study reveals that individuals have a poor understanding of the cost of driving. Looking at ubiquitous technologies we believe that there is a great potential for saving energy. Information can help to reduce the need for driving. We make concrete suggestions of how user interfaces in the car can make the user aware of the potential of reducing fuel consumption.

An Interactive Knocking Floor

Many interactive floor systems have been introduced with increasing focus on the floor. However, most studies considered the floor as a large screen that we can walk over, and few studies have focused on the relationship between a person and the floor. In this paper, we introduce an interactive knocking floor as an ambient invisible floor interface for public space. This floor tile exerts impact against user's feet stepping on the floor. To show the performance of our device, we measured working sound volume and conducted user tests for our device together with a beeper device and a light device in both the walking case and the standing case.

Detour Scheme for Improving the Energy Efficiency and Fairness in Wireless Sensor Networks

Wireless Sensor Network (WSN) is comprised of sensors which are not recharged. Sensors consume their energy for transmitting sensing data and relaying it. The number of sensors run out of energy increases as the time goes by. A 'void' and 'hole' are the main cause of shortening the lifetime of wireless sensor networks. In this paper, we propose a novel detour scheme, which improves the energy efficiency and fairness in wireless sensor networks with a 'hole'. The proposed scheme selects a detour anchor point based on the distance between a source and a sensor node which detects a hole. If the distance is short, the anchor point uses a big detour angle and lengthens the distance from itself to the sensor node. Meanwhile, if the distance is long, it uses a small angle and shortens the distance from itself to the sensor node. The proposed detour scheme can solve the problem of a hole being enlarged. Moreover, it can improve the energy efficiency since it balances traffic load by selecting a proper anchor. The transmission delay fairness is provided since the proposed scheme uses a different weight according to the distance from a source when it selects a detour anchor point.

Using Personalization to support Privacy in Ubiquitous Systems

In developing pervasive computing technologies that are acceptable to the end user, it is essential to take account of user needs and preferences to personalize decision making within such a system. One important area where they may be used to improve the behaviour of pervasive systems is in dealing with the problems of maintaining user privacy. User preferences can be used both in determining what information can be released about the user and in selecting a virtual identity to hide the real identity of the user. By building up a flexible context-aware set of user preferences that can be used to assist in taking these decisions, one can increase the degree to which this can be handled automatically, and improve acceptability by the user. This paper outlines the approach and describes how it is being implemented in the Daidalos pervasive system.

Rich Tactile Output for Notification on Mobile Phones

In this paper we assess the potential of rich tactile notifications on mobile phones. Many mobile phone users use vibration output for various types of notification on their phone. Currently, tactile output in phones is limited to one single actuator that can potentially present patterns based on different vibration intensities over time. To explore the possible design space, we conducted experiments with up to 6 actuators included in a prototype phone. The dimensions of the resulting design space are comprised of the location of the active actuators, the intensity of the vibration, and the variation of these parameters over time. Based on several user studies and interviews, we suggest design guidelines for rich tactile output facilitated by several distinct actuators. We show where vibration motors should optimally be placed and that different information can be communicated by producing different patterns of vibration output using a small number of actuators.

Creating Logfiles and Click Streams for Advertisements in Physical Space

Poster advertisement has a long tradition and is transform-ing rapidly into digital media. In this paper we provide an overview of how sensing can be used to create online and up to date information about potential viewers. We assess what application domains can benefit from continuous monitoring of visitors. As measuring with simple sensors is inherently error prone we suggest the notion of comparative advertising power which compares the num-ber of potential viewers in different locations. We address user acceptance and privacy concerns and show technical mechanism to increase privacy.

ECOISLAND: A System for Persuading Users to Reduce CO2 Emissions

Significant portions of the carbon dioxide emissions that have been shown to cause global warming are due to household energy consumption and traffic. EcoIsland is a computer system aimed at persuading and assisting individual families in changing their lifestyle patterns in a way that reduces CO2 emissions. The system builds on our earlier work on persuasive ubiquitous computing applications, and applies ideas from behaviorism, social psychology and emissions trading to motivate changes in users behavior. In this paper, we briefly describe the concept and the theories behind it, and provide preliminary results from a user study measuring its effectiveness.

SpinTrack: Spinning Infrastructure Nodes for Precise Indoor Localization

We propose SpinTrack, a localization system enabling accurate tracking of both stationary and moving targets in the indoor environments. SpinTrack integrated two complementary methods, SpinLoc (for tracking stationary targets) and dTrack (for tracking moving targets) into one localization system. We propose how to eliminate the transitional overhead between the two methods in order to optimize the overall system performance.

Analysis and Comparison between War Driving and War Walking in Metropolitan WiFi Radio Maps

War-driving is currently the most widely adopted method for building large-scale radio maps in metropolitan Wi-Fi localization. Although the human effort cost for war driving is smaller than that of war walking, its positional accuracy is also lower than that of war walking. This work compares radio maps built from war driving and walking and analyzes how selective map characteristics affect positional accuracy.

Adding Manga-like Effects to Your Presentation with an Acceleration Sensor

Nonverbal information is significant in communication. We propose an easy and effective way to make your presentation impressive by emphasizing nonverbal information. Unconscious body movement is detected by a wireless acceleration sensor (Wii Remote) held in the user's hand. And the emphasis is overlaid on the presentation slide with manga-like exaggeration. For example, dynamic pointing motion which appears when the presenter emphasizes his/her talk raises concentration lines on the screen. Besides, other features to support the presentation are implemented. The presenter can concentrate on his/her talk with the support and express his/her emotion better. The system is especially effective in rooms with larger screens.

MOBBMap: Public Authoring Around Public Transit

Based on an ethnographic study of urban mobilities around Bangkok, we present an SMS-based, user annotated, public transit route-finding system called MOBBMap (Mobile Open Bangkok Bus Map). MOBBMap enables public authoring and attempts to provide a forum for residents of Bangkok to demonstrate local expertise, share their images of the city, point out mobile dwelling resources, as well as simply find out about bus routes. Development is currently underway.

Social Interaction Detection and Browsing in Meetings

This paper proposes a multimodal approach for detection and browsing of human social interactions in meetings. It first presents a method to recognize interactions based on a variety of contexts, including head gestures, attention from others, speech tone, speaking time, interaction occasion (spontaneous or reactive), and information about the previous interaction. Then, a graphical user interface called MMBrowser is presented for interaction browsing. Experimental results verified the proposed approach.