1. Personal Server Model for Real-Space Networking Shirou WAKAYAMA shirou@sfc.keio.ac.jp KEIO University Japan.

2. Motivation Attain Computing Environment using information in Real-Space Active Bat (Location-aware system using RF) GUIDE (Location-aware system using Wireless LAN) Oxygen (Location-sensing and using embedded computers around the user) Can not utilize various sensors in each system → Limited system availability and sensor variation We want to use various sensors in more Wide-area

3. Sample Applications Autonomous information obtainment current location name of the person standing before the user Autonomous health-care support Call ambulance automatically, when user ’ s physical condition change suddenly Autonomous diary generation Log user ’ s daily activity automatically

4. Assumed Environment Pervasive/Ubiquitous Computing environment Users can be unaware of computers and devices Surrounding environment should care Centralization to Distribution! Users move in wide-area Outside of lab, campus, town, everywhere Various public sensors are setup everywhere Sensors differ depending on places Sensors nearby user will sense and inform The Internet is used to connect sensors each other

5. Problem statement Sensor information Management Model is required No common interfaces between each research/sensors Each research has own Sensor information managers and models No work has done to provide interoperability No model to make use of various sensors while moving anywhere No overarching model

6. Model Requirements Sensor information be obtained anywhere Location Independent Various sensor-devices be utilized Sensor-type Independent Information be shared between people requested Only authorized people could request

7. Existing model consideration 1. Aggregate/Manage all sensor information to one Central Server Less scalability 2. Manage sensor information by servers corresponding to each sensor As users move, surrounding sensors will change accordingly Other users may not know where to request sensor information of specific user

8. Personal Server Model(1/2) Manage Sensor information by servers corresponding to each user Manager can be set anywhere on the Internet Manager can aggregate/manage information from various sensors bound to each user Other users can request sensor information by user specified Location/Sensor-type Independency in Pervasive Computing environment provided

9. Personal Server Model(2/2) B Real-Space Virtual-Space Personal Server = Sensor information manager Globally Unique ID ‘ GUID ’ A ABC C Middle Sensor Layer Sensors

10. Experimentation 2002 Sep., with 274 people, 4days 3night In the WIDE Project meeting Sensor information Location-information from RF-ID and IP address Temperature information of electric hot-water pot Information service via Web interface Locate a person List people in a room specified Temperature information of pot

11. PS System overview HTTPDPS A B PS = Personal Server Other Application HTTP SQL RF-IDIP Address Pot Users

12. Location Sensing by RF-ID RF-tag Sensor Send Sensor Information to Personal Server of GUID owner To Personal Server Antenna Ethernet RS-232C Antenna Cable Sensor device(1/3) RF Base Stations All User had RF-tag, that Send GUID through RF.

13. RF Base Station Coverage

14. Sensing by IP Address IP address for location sensing IPv6 segment(prefix) differ per room DB has binding of prefix and room name Location information of Note PC per room can be provided Sensor device(2/3)

15. Pot (Temperature Information) Micro Node Temperature Sensor Sensor Device(3/3)

16. Locating a person Service(1/2) Locate your friend Find out from GUID Find out from Name Find out from Organization

17. Listing people in a room specified Click the room, List people Service(2/2)

18. Sensed Movement BOF1 BOF2 BOF3 LOBBY1 LOBBY2 LOBBY3 LOBBY4 PLENARY1 PLENARY2 NOC Actual Movement Sensed Movement 0:0012:0024:00

19. Location Information from RF-ID and IP addr. BOF1 BOF2 BOF3 LOBBY1 LOBBY2 LOBBY3 LOBBY4 PLENARY1 PLENARY2 NOC 0:0012:0024:00 IP addr RFID Divide One large room to Two

20. Future Work Test scalability of Personal Server Model Not evaluated about Scalability with large number of Personal Servers and users Not evaluated about Scalability of mechanism that resolves IP address of Personal Server from GUID Privacy Protection Secure communication between Sensor and Personal Server Secure user authentication

21. Conclusion Personal Server Model Personal Server manage sensor information per user Personal Server Model realizes Location independency Sensor-type independency Pervasive Computing environment Sharing information between people Experimentation Multiple sensor devices used for information acquisition Web interface used Personal Server Model can apply to the everyday use