PicoRadio The State of the Union

1
PicoRadioThe State of the Union

• Profs Rabaey, Sangiovanni-Vincentelli,
  Ramchandran, Wright, Brodersen,
• Katz, Pister
• PicoRadio, TCI, and Protocol Groups

2
PicoRadio Charter

• Develop meso-scale radios for ubiquitous
  wireless data acquisition that minimize
  power/energy dissipation
• Minimize energy (lt10 pJ/(correct) bit) for
  energy-limited source
• Minimize power (lt 1 mW) for power-limited source
  (e.g. based on energy scavenging)
• By using the following strategies
• self-configuring networks
• fluid trade-off between communication and
  computation
• aggressive low-energy architectures and circuits

3
PicoNode Goals and Milestones (June Retreat)

• June-Oct 99
• Target specs
• Definition of application demos
• Oct 1999 - Oct 2000
• HW SW Simulator
• RF Testbed
• Start looking at applications
• Design environment fixed
• HW - SW rendezvous

• Oct 2000 - Oct 2001
• Piconode Chip
• Handoff of Piconodes
• Oct 2001 - Oct 2002
• Chip Network Application

4
PicoRadio Applications and Specs
The Obvious Choice -The Smart Home and Network
Appliances
5
The Interactive Museum
6
Industrial Building Environment Management

• Task/ambient conditioning systems allow thermal
  condition in small, localized zones (e.g.
  work-stations) to be individually controlled by
  building occupants
• Requires dense network of sensor/monitor nodes
• Wireless infrastructure provides flexibility in
  composition and topology
• Joint research proposal CBE/BSAC/BWRC

7
Interacting Toys
To bring about the existence of a broad range of
plug-and-play interoperable PC-connected and
networked toys by defining common requirements on
untethered communication infrastructure between
toys and PCs that enable play nearby the PC or
anywhere in and around the home.

• Stated requirements
• Plug-and-play
• Interoperable
• Networked - preferably with pc
• HomeRF compatible

8
System Requirements and Constraints

• Constraints
• Stringent requirements on size (lt 10 cm3) and
  cost (lt 10 ) per node
• Energy consumption per node must be kept to an
  absolute minimum time-between-recharging gt
  years, or P lt 0.5 mW
• System should be self-assembling and operation
  should be foolproof
• Specifications (from Exploratorium scenario)
• Large numbers of nodes (between 0.05 and 1
  nodes/m2)
• Limited operation range of network (maximum
  50-100 m)
• Low data rates (1 - 10 kbit/sec) (should
  potentially support non-latency critical voice
  channel)
• Spatial capacity (or density) in bits/sec/m2
  crucial design parameter

9
Removing Dependencies in Research

• PicoRadio addresses research issues in many
  disciplines
• application, protocol and network, system and
  chip architecture, digital low-energy circuits,
  RF
• Interdependencies hinder progress
• PicoRadio project temporarily partitioned into
  two (independent) projects to break dependencies

10
TCI An Exercise in Integrated Wireless Node
Design
Basestation

• Known and tested specification of limited
  complexity allows focus on architectural
  implementation methodology
• Two-chip implementation leverages separates
  between analog (RF) and digital design concerns
• Duration of exercise 1 year (summer 00)
• Close cooperation with GSRC(as driver and user)

Mobiles
Up to 20 users per cell _at_ 64 kbit/sec per
link TDMA selected as MAC protocol
11
Target The Integrated PicoNode
12
Two-Chip Intercom (TCI)
Program-mable logic
Software running on processor
Custom analog circuitry
Mixed analog/ digital
Fixed logic
Protocol
ADC
Digital Baseband processing
Analog RF
DAC
Chip 2
Chip 1
13
System Design and Optimization Hierarchy
Network level
Constraints
Node level
14
Proposed Protocol Design FlowJune Retreat (Marco
Sgroi)
System Spec
Input language (ECL,SDL)
Model of Computation CFSMs
Co-simulation Formal Verification
Refinement
Stateflow/Simulink
Implementation Hw (VHDL) / Sw (C)
15
Protocol Design Flow
16
Describing The Protocol Stack

• Description in Cadence VCC MOC CFSM
• Other models Polis, SDL, JavaTime

17
Digital Baseband
Stage 1 floating point blocksStage 2 fixed
point blocks
Stage 1 Model components in MATHWORKS tools
(Simulink/Stateflow) at appropriate
granularities Stage 2 Convert Simulink
structural blocks (manually) to fixed point /
bit-true Stage 3 Map to HW and/or SW (automatic
synthesis path Simulink to HW - Simulink to
Software)
18
Targeted Implementation Platform
Memory Sub-system
Tensilica Embedded Proc.
Sonics Backplane
Programmable Protocol Stack
ConfigurableLogic (Physical Layer)
Baseband Processing
Benefit Build library of computational and
networking modules (and models)
19
Protocol Implementation Challenges
TDMA link and MAC alternative implementations
Intercom transport and application layers on
Xtensa processor (from SDL generated code)

• Observations
• Major differentiation in functionality
  between different levels of stack
• Tremendous opportunity for optimization at
  description and implementation level

20
System Optimization Hierarchy
Network level
Constraints
Node level
Think Energy!
21
PicoRadio Energy OptimizationThe Cost of
Communication
Assumes R-4 loss due to ground wave (_at_ 1 GHz)
90dBm
90dBm
100 Kbps
50dBm
50dBm
Transmit Power
Transceiver Power
10dBm
10dBm
-30dBm
-30dBm
-70dBm
-70dBm
1m
10m
100m
1Km
10Km
Distance
22
Communicating over Long DistancesMulti-hop
Networks

• Example
• 1 hop over 50 m
• 1.25 nJ/bit
• 5 hops of 10 m each
• 5 ? 2 pJ/bit 10 pJ/bit
• Multi-hop reduces transmission energy by 125!
  (ignoring overhead and cost of retransmissions)

Source
Dest
23
Energy-Optimizing Multi-hop Networks
fs
where
and
is the ceiling function
A constant relating the energy required to
transmit a bit successfully for a given set of
parameters.
b
A constant relating the computational cost for
receiving the bit
a
Optimal number of hops needed for free space path
loss.
24
Process Model
Node Model
Network Model
Analysis Viewer
OPNET Network Simulator
25
ExampleTable-Driven Network Routing
Network maintains routing information proactively

• Assumptions
• Max nodes represented in single update 50
• Checkerboard Placement
• Mobiles Enter Stable Network Simultaneously
• No Packet Loss
• Num of Nodes 50 - 55
• Update to Neighbors Only
• DSVD Routing

Time to disseminate New Info
Other options Source Initiated or Reactive
Routing
Additional Updates Required
26
Joined Networking-Positioning
1 error

• Ubiquitous radio networks offer accurate
  localization with minimal overhead
• Potential to substantially reduce networking
  overhead!

27
Adding the Activity Factor

• Energy activity cost distancen
• Activity in sensor networks is low and random
• Major opportunity for power management
• Best addressed at the media-access (MAC) layer of
  the protocol stack
• Non-active nodes should be in sleep mode as much
  as possible
• Media-access should be such that collisions and
  retransmissions are minimized

28
Energy-Efficient Media Access

• Example Collision-sense multiple access (CSMA)
  with overlayed locally-synchronized TDMA framing

RX/TX in sleep mode
time
Sender 1
CSMA
Sender 2
Other approaches purely asynchronous (reactive)
29
Mostly-sleepy radio design

• Harmonized design between air interface, radio
  architecture and higher level protocols essential
  to maximize power-down
• Receiver should be dozing most of the time, but
  needs to keep a low-power ear to the channel
• Reactive approach normal reception chain gets
  activated up by wake-up signal of transmitter.
  Requires dual radio architecture.
• Proactive or scheduled wake-up approach - easiest
  from radio perspective, but inefficient in ad-hoc
  dynamic networks
• Simulated asynchronous receiving nodes transmit
  Im awake signals
• Radio adaptivity to meet variant channel
  conditions and application requirements minimize
  transmission or listening overhead.
• Configurable radio parameters processing gain,
  A/D resolution, transmission energy level,
  modulation scheme, coding, etc. Trade off with
  higher level parameters packet length,
  retransmission scheme, etc. in order to optimize
  power performance.

30
The Holy Grail Energy Scavenging
Energy Sources
Integrated micro-vibrator provides 10-100 mW of
free power (equivalent to 2340 free DSP
operations/sec) Amirtharajah Chandrakasan,
DISPS99
31
A Testbed for PicoRadio

• Flexible platform for experimentation in
  networking and protocol
• Projected size 3x4x2
• Multiple radio modules Proxim, Bluetooth
• Collection of sensor and monitor cards
• Available soon!

Xilinx
Flash ROM 256Kx16
Board to Board Connectors Test Headers
Strong Arm
32
PicoRadio Design Challenges
PicoNode Architecture Design
Positioning Network Architecture
Performance Analysis
Energy Constraints
Analytical Analysis
Use Cases
33
What have we accomplished?

• Clear perspective on applications and
  specifications of PicoRadio
• TCI established and executed comprehensive design
  and exploration flow for protocol design.
• Hopefully leading to full implementation by
  summer 2000
• Gained clear understanding and knowledge of
  issues in wireless ad-hoc networking
• Established analysis methodology
• Merged networking/localization approach very
  promising!
• Some successes at the circuit and architecture
  front
• Two papers to be presented at ISSCC00 (DVS
  Microprocessor and Maia reconfigurable processor)
• Low-energy FPGA selected for Low-Power Design
  Award at Computer Elements Workshop

34
Where are we going (next 6 months)?

• Exploratorium exhibit in BWRC - simple sensor
  network implemented using PicoRadio testbed and
  Bluetooth radios
• TCI digital chip close to completion
• Draft proposal for PicoRadio network and protocol
  stack
• Proposal for low-energy implementation fabrics
  for protocol stack