PicoRadio Poster Introductions

1
PicoRadio Poster Introductions
Development Platform (4) Energy Scavengy
(1) Low Energy Triangulation (1) Low Energy
Physical Layer (1) Protocols Low Energy
(3) Design Methodology (3) Formal
Verification (2) Architecture Exploration
(2) Implementation (3)

• BWRC Retreat 6/2/00

2
The PicoRadio Development Platform

• System Overview
• What it is a collection of nodes forming a
  highly configurable development platform for
  PicoRadio prototyping.
• What it provides a target architecture, an
  integrated compilation and synthesis environment,
  a library of system code and logic, and a network
  
• Architecture flexible I/O, microprocessor,
  programmable logic
• Deployment small remote units in a distributed
  system
• Companion Posters
• System hardware
• Software and logic support

Susan Mellers Fred Burghardt
3
The First PicoNode
RESET
Serial Port
DIGITAL
Board to Board Connectors
BATTERY
POWER
On/Off Switch
RADIO ADAPTER
RADIO
Susan Mellers Fred Burghardt
4
Software and Logic Support Basic Elements

• Design environment
• Compose and compile ARM applications
• Build ARM executable for embedded operation
• Debug ARM executable
• Design and synthesize configurable logic for
  Xilinx
• Simulate Xilinx design
• Library of system code and logic (glue)
• Operating system and device drivers
• Virtual channel abstraction
• User Application Programming Interfaces (APIs)
• Read/write API into Xilinx concept of ports
• Interrupt support for Xilinx ? ARM asynchronous
  events

Susan Mellers Fred Burghardt
5
Designing Electromechanical Devices

• Summarize two challenges in designing
  electromechanical devices
• Communication difficult between electrical,
  mechanical design environments
• Casing must conform to device requirements, and
  manufacturing requirements (in this case,
  injection molding)

Methodology used to design PicoRadio casing.
Dan Odell Michael Montero
6
Energy Scavenging

• Goal provide enough energy to power wireless
  sensor node
• In part, or in whole
• Preliminary investigations completed
• Solar energy, acoustic waves, mechanical
  vibrations, nuclear energy, and the human body
• Focus on mechanical vibrations
• Designed capacitive MEMS generators
• Convert mechanical vibrations to stored
  electrical energy
• Simulations indicate 10-100s mW can be generated
• Future plans
• Test MEMS devices
• Model, building inductive and piezo-electric
  converters
• Fabricate, test integrated solar cell / micro
  battery structure

Shad Roundy
7
Energy-Efficient Triangulation in a PicoRadio
Network

• Goal
• Low-power, accurate triangulation in an ad-hoc
  network
• Primary Obstacles
• Potentially all nodes at unknown locationNo
  beacons, etc.
• ?50 error in RSSI measurements used in
  triangulation calculations
• Solutions Approach
• Allow for expensive start-up, optimize
  steady-state updates
• Use CORDIC algorithm to achieve efficient updates
  to linear equations

Chris Savarese
8
Physical Layer Digital Baseband

• Timing recovery circuitry accounts for more than
  half the power of the TCI receiver.
• This poster describes
• Digital timing recovery algorithm for TCI
• Hardware implementation approach
• Improvements to make for PicoRadio

PicoRadio requires 100x improvement over TCI!
M. Josie Ammer
9
PicoRadio Networking

• Identified type of connections required in ad-hoc
  sensor networks
• Identified suitable routing protocol for each
  type of connection
• Defined a hybrid geographical, reactive,
  diffusion based routing

Rahul Shah Danny Patel
10
Low Power MAC Strategy

• Key features of PicoNode MAC
• Fully distributed protocol, no central control or
  synchronization robustness, scalability
  flexibility.
• Multiple channels to reduce collision rate and
  signaling overhead minimize of bits through
  antenna.
• Power down the radio AMAP adaptive to local and
  relay traffic.
• Broadcast support
• Separate passive wake-up radio
• Our heretic approach vertical design, across
  layer boundaries, to optimize from system
  perspective.
• Opnet simulation model, VCC implementation flow

Chunlong Guo
11
Lower Power MAC Design for Pico Radio Project

• Multi-dimensional random access
• Only wakes up for transmission/reception
• Learn from your neighbors
• Multicast for better power efficiency

Original sender
Node for retransmission
Charlie Zhong
12
Extending POLIS with Data Types

• An implementation for extending the POLIS
  co-design environment
• Support for user-defined data-types
• Modifications to the POLIS software hierarchy
• High level (input formats)
• Intermediate level (SHIFT format)
• Low level (HW/SW synthesis)

Arvind Thirunarayanan
13
Design Methodology

• Design performed using POLIS
• Modules written in Esterel
• E-CFSM model of computation used to represent
  both control and dataflow, and is unbiased to HW
  or SW
• System can be simulated, verified, and trade-offs
  evaluated
• Provides a direct path to implementation
• Since we are designing a layered protocol, we can
  complete the design from the top down.
• At each layer, make certain assumptions about
  what lower layers are providing, meet certain
  requirements higher layers place on us
• Can design each layer, simulate and verify
  correctness before moving on

Jason Shamberger
14
Metropolis
Roberto Passerone Marco Sgroi
15
Formal Verification of TCI

• System satisfies certain properties?
• System described in some formal mathematical
  languages (e.g. Esterel)
• Properties written in some formal logic or formal
  model (e.g. Esterel)
• Approach used Property Verification
• Invariant (only one remote can send voice data at
  any time slot)
• Response (if a remote sends a request to the base
  station, then eventually there is an
  acknowledgement)
• deadlock freedom
• Mocha Modularity in Model Checking (UCB)
• model checker developed in Prof. Thomas A.
  Henzingers group
• TCI Transport Layer verification
• 5 Properties checked 2 failed, both response
  properties

Freddy Mang
16
FRESCO Formal Real-Time Software Components

• Modeling formalism based on modularity and
  hierarchy
• Refine design of a complex system, preserving
  formal semantics
• Make design amenable to verification
• Top level components with serial and parallel
  composition
• Software implementation processes running on top
  of RTOS
• Refinement using compositional and
  assume-guarantee rules
• Application model and verify part of PicoRadio
  design flow

Marius Minea
17
Functional Profiling

• Goal Identify critical operations in layers of
  wireless protocols
• Key Recognize critical operations implement
  them efficiently
• Network processor community
• Use functional profiling to design network
  processor architectures
• critical operations parsing, searching, packet
  modifying, re-assembly etc.
• Preliminary results from functional profiling of
  wireless application
• using Opnet (AODV and DSDV protocols)
• Network layer critical operations similar to
  network processor operations
• Searching, packet processing operations consume
  significant cycles
• More experiments needed to draw conclusive
  results.

Suet Fei Li
18
Architecture Modeling and Exploration

• Tensilica Xtensa Processor
• Modeling for VCC code estimation
• Improving model accuracy
• Sonics SiliconBackplane Interconnect
• Partitioning TCI into separate cores
• Modeling SiliconBackplane in VCC
• TCI System Architecture
• Proposed TCI architecture
• Communication refinement issues in VCC

Mike Sheets Vandana Phrabu
19
Protocol Implementation From VCC to HDL

• Method for translating STDs written in VCC into
  Verilog
• Verilog template
• Recipe
• Optimizations
• Applied this method to TCI protocol
• 600 lines of C code 75 states gt 6400 lines of
  Verilog code
• Compared to existing automated solutions
• Area is ¼ from the best automated solution

Julio Silva Tim Tuan
20
Protocol Implementation Mapping to Hardware

• This poster presents the hardware implementation
  methodology of TCI, including
• RTL simulation
• Hardware performance estimation
• Partitioning system onto separate platforms
• Low-energy FPGA implementation methodology
• Integration challenges

Julio Silva and Tim Tuan
21
Low-Energy Reconfigurable Fabric for Protocols

• This poster reviews the preliminary work towards
  the design of a reconfigurable fabric for
  implementing the Pico Radio protocol stack
• Compare existing reconfigurable fabrics
• Draw early conclusions
• Look ahead to next step

Tim Tuan