The Universal Content Router

1
The Universal Content Router

• Jana van Greunen
• The UCR is part of the ZUMA project
• Students C. Baker, Y. Markovsky, A. Schultz, A.
  Krasnov, S. Chen
• Professors Jan Rabaey, John Wawrzynek and Adam
  Wolisz

2
Introduction

• Trends in the information age
• Information anywhere, anytime
• Small, multi-function devices store content
  must be syncd
• Host of networking technologies

• Problems
• Difficult to setup devices (esp. across
  categories TIVO cellular)
• Isolated clusters of connectivity
• Multiple user interfaces, with no continuous user
  experience

3
Project Background
Converged Media-Sensor Networks (CoMSen)
Media Networks Resident Audio/Video Content,
Broadband (voice/data), Rendering
Devices High-Bandwidth, Compute and Memory Capable
Sensor Networks Sensing, Control,
Actuation, Low-Bandwidth, Compute and Memory
Limited
The UCR broadens CoMSens scope, with an
infrastructure focus
4
UCR Objective

• To render a piece of content on any device
• Guarantee the quality of the user experience
• Allow plug play operation of different hardware

UCR
5
Future Intelligent Home

• New application categories are emerging
• Energy management
• Environment control
• Security Surveillance
• Health Care

6
The Portability Scalability Challenge

• New devices are entering the home environment at
  an ever increasing rate
• Standards are proliferating interconnect,
  recording and playback, display
• Devices do not necessarily interconnect easily

The Home Network as the new Tower of Babel
7
ZUMA Home Platform

• Zero-Configuration minimal need for device
  configuration
• Universality ability to connect any device to
  any other device
• Multi-User Optimality optimize user experience,
  with multiple users and simultaneous tasks
• Adaptability change according to users desires,
  location, and integrate new devices

8
Two-pronged Approach
More info - ZUMA A Platform for Smart-Home
Environments Baker, Markovsky
Top Down
Programming abstraction Resource
abstraction Ambient OS
Routing mapping Resource management Seamless
connectivity
Bottom Up
9
Challenges Requirements

• Physical layer must be flexible configurable
  to support future protocols
• Protocol conversion between standards such as
  802.11, 802.15.4, Bluetooth, etc.
• Routing Dynamic routing based on devices
  capabilities
• Presentation conversion devices expect different
  data formats data rates (trans-coding problem)
• Determine what type of trans-coding is needed
• Real-time constraints
• Discovery Uniform way to discover integrate
  new devices

10
Bottom-up Solution Universality

• Stovepipe (develop a single standard)
• Not feasible in free-market with existing solns
  
• Apps have different requirements (power,
  bandwidth)
• Flexible peripherals (Cell phone model)
• Not economically viable
• Hard to incorporate new standards (not future
  proof)
• Overlay centralized infrastructure
• Acts as NxN switch
• Easier for management upgrade
• Easy to deploy

11
UCR Structure
Configurable fabric / mix
pre-config blocks with
configurable
inter-connect
Ambient OS Application Management Backplane
Configuration Allocation
Phy
Link
App

Flexible SDR front-end
12
UCR Operation
1) Connecting a bluetooth PDA to a laptop (w/o
bluetooth)
2) Transcoding multimedia (MPEG to JPEG on wifi
network
Ambient OS Application Management Backplane
Configuration Allocation

13
Future Physical Layer Connectivity

• The future is WIRELESS
• Future-proof radio (SDR)
• Scheme for detecting decoding a new protocol
• Continuously scan a wide range of frequencies
• Upon detecting a new signal
• Match new frequency profile with all known
  profiles (Info can be stored in an online
  database)
• Implements the closest match decode the signal
• If the decoded signal is nonsense, try next
  closest match

14
Bandwidth Considerations

• Bandwidth Requirements
• Unlicensed channels (2.4 5.2-8GHz) are
  insufficient
• Solution licensed spectrum in TV Bands
  (Cognitive Radio)
• How much can we gain? Approx 55 unused channels /
  area!!
• HomeGateway/UCR will be a driver for CR
• Infrastructure makes sense for a CR system

15
Computation Requirements

• Protocol conversion
• Computation mostly at PHY layer
• Standards vary OFDM steps include frequency
  offset correction, a 64-point FFT, channel
  equalization and QAM demodulation.

16
Prototype UCR
Universal Content Router
BEE2
802.11pre-n Bridge
Control FPGA
xx.yy.zz.1
eth0
100Mb/s Ethernet
eth1-4
Slave FPGA2
Slave FPGA5
Slave FPGA4
Slave FPGA1
xx.yy.zz.2
xx.yy.zz.4
xx.yy.zz.5
xx.yy.zz.3

• All devices use Internet Protocol
• Data RTP (Real Time Protocol) encapsulated in
  UDP
• Control Home Gateway Control Protocol (HGCP)
  encapsulated in TCP

• Control FPGA runs Linux
• A router between the Slave FPGAs and the outside
  world.
• Four Slave FPGAs
• Run-time configured with required trans-coder

17
Prototype Demonstration

• Scenario
• Home multimedia-watching application
• Video follows user around the house
• Setup
• House contains two rooms with displays
• Sensor network detects the users location
• Displays require different media formats and
  trans-coding happens on the fly

18
UCR Demo Hardware
Content Storage
Universal Content Router
Presence Sensor
Presence Sensor
Speakers
19
Conclusion

• Our vision for the intelligent home
• Baker, Markovsky, et. al A Platform for
  Smart-Home Environments IE06, 2006
• Van Greunen, Markovsky et. al. A Platform for
  Smart-Home Environments The case for
  Infrastructure IE06, 2006
• Creating a Testbed in BWRC
• Media processing reconfigurable platforms
• Cognitive radios
• Sensor networks
• Distributed resource management
• Key challenges will be discussed in breakout
  session C