Wireless Communication Wireless Communication

on Wearable Systemson Wearable SystemsCORECO I, WEMS II + IIICORECO I, WEMS II + III

Jan Beutel, Computer Engineering and Networks Lab

Mathias Stäger, Holger Junker, Electronics Lab

December 4, 2002

Po

lyp

roje

ct W

ear

abl

e C

om

put

ing

, E

TH

Zü

rich

, Dec

emb

er 4

, 200

2

The Body Area Networking DifferenceThe Body Area Networking Difference

• Distributed Configurable Computing Platform

– heterogeneous components

– communication centric

– low power

– varying configurations and requirements

– many components (~20…50)

Display

Context Sensor Array Camera, light,

microphone, GPS

Distributed Reconfigurable

Computer

Body Area NetworkWired and wireless

Access Networking Bluetooth, WLAN

variants, GSM,UMTS, Thuraya

Sensors

Interaction with ubiquitous appliances

Audio

Po

lyp

roje

ct W

ear

abl

e C

om

put

ing

, E

TH

Zü

rich

, Dec

emb

er 4

, 200

2

CompetitorsCompetitors

• Wearable Systems– single garment, wired-up solutions

MIThril (DeVaul)– custom box-type computer (Starner)

• Sensor Networks– low bit-rate SOC transceivers (Rabaey)– COTS sensor networks (Pister et al)

• Power Management– dynamic power saving states

(De Micheli, Gupta)– low power frontends (Enz, Meng)

Po

lyp

roje

ct W

ear

abl

e C

om

put

ing

, E

TH

Zü

rich

, Dec

emb

er 4

, 200

2

Our Research IssuesOur Research Issues

• Scalable low power networking on Wearable Systems

– optimal use of resources

– low power body area networking transceivers

– fast prototyping for the implementation of real life scenarios

• Design parameters from the network view– network topology– spatial capacity– link data rates– latency/burstiness– transceiver architectures– protocol features

Po

lyp

roje

ct W

ear

abl

e C

om

put

ing

, E

TH

Zü

rich

, Dec

emb

er 4

, 200

2

Investigation of Standard ConceptsInvestigation of Standard Concepts

• Why standardized wireless devices?– operational components available today for prototypes– foundation of methodical approach to distributed systems

• Goals– knowledge to select the right hard/software architectures– optimal duty cycle performance dependant on the application– future replacement by custom components

• Our contribution– characterization and benchmarking of existing wireless communication

devices, protocols and transmission schemes – results are used in the modeling of communication channels for Design

Space Exploration of Wearable Systems [Anliker et al, submitted to TOC]– implementation of a Bluetooth protocol stack on Linux and uC

Po

lyp

roje

ct W

ear

abl

e C

om

put

ing

, E

TH

Zü

rich

, Dec

emb

er 4

, 200

2

Communication ModelCommunication Model

• The Problem optimal device configuration for each linkPower Consumption and Delay

• Assumptions– multiple periodic inputs– deadline associated to data– four operating states: standby, idle, transmit, receive

continuousduty cycle 1

burstduty cycle 2/3

Po

lyp

roje

ct W

ear

abl

e C

om

put

ing

, E

TH

Zü

rich

, Dec

emb

er 4

, 200

2

BTnodes – Bluetooth Smart NodesBTnodes – Bluetooth Smart Nodes

• Programmable networking node for

fast prototyping

– 8-Bit RISC CPU, (max. 8 MIPS @ 8 MHz)– 128 k Flash, 64 k SRAM, 4k EEPROM– generic sensor interfaces– power and frequency management– Bluetooth with integrated antenna

– idle @7.3 MHz, 3.3V <0.5 mW– active @7.3 MHz, 3.3V 150 mW

• Status– initial sw kit, drivers and demo

applications available– current deployment ~ 200 (12

research groups worldwide)– [Beutel et al, submitted to MobiSys]

Po

lyp

roje

ct W

ear

abl

e C

om

put

ing

, E

TH

Zü

rich

, Dec

emb

er 4

, 200

2

Integrated Network Protocols using Integrated Network Protocols using BTnodesBTnodes

• Bluetooth Multihop Prototype– integrated scalable application

protocol– based on Dynamic Source

Routing (CMU)– routing across piconet borders

to support >8 nodes

• Status– first implementation on

BTnodes available

– integration with the Physical Activity Detection Network PADnet (Demo)

Po

lyp

roje

ct W

ear

abl

e C

om

put

ing

, E

TH

Zü

rich

, Dec

emb

er 4

, 200

2

Comparison Transceiver ArchitecturesComparison Transceiver Architectures

• Goal– dedicated architecture for on-body sensor network

– short distance (50 cm)– low bit rate (0.1 ~ 1 kbit/s)

– low power consumption (100 W targeted)

• Comparison of– traditional far field architectures– near field systems

– magnetic induction– capacitive coupling

– ultra wideband architecture

• UWB as promising new candidate– simplicity of implementation– relatively new research field

Po

lyp

roje

ct W

ear

abl

e C

om

put

ing

, E

TH

Zü

rich

, Dec

emb

er 4

, 200

2

The Road so FarThe Road so Far

• Interfacing within the Polyproject – modeling of communication systems

► Design Space Exploration CORECO IV– BTnode

► user activity network and reconfigurable computing CORECO I + IV

• In two years…– a refined communication model

– on-line tradeoff and application of different interface types

– UWB channel characterized and suitable architecture implemented

– integration into the demonstrator platform