High Data Rate Wireless and status

1
High Data Rate Wireless (and status)
BWRC Summer Retreat 2003
Bob Brodersen Dept. of EECS Univ. of
Calif. Berkeley
2
FCC - Unlicensed Spectra
3
Lower Power Wider Spectrum

• Challenges
• Intelligent / Adaptive algorithms
• Better interference management
• More cooperation between users

4
Shorter Innovation Cycle

• According to FCC, one of the six goals for the
  next 5 years (Strategy Plan FY 2003-FY 2008)
• Encourage the highest and best use of spectrum
  domestically and internationally in order to
  encourage the growth and rapid deployment of
  innovative and efficient communications
  technologies and services.
• Challenges
• Flexible architecture

5
Dimensions for Spectrum Use

• Geographic space
• Power
• Time (when, how long)
• Frequency (where, bandwidth)
• Antenna array
• Users

6
Summary of Projects

• Individual Dimensions
• Ultra-wideband
• 60 GHz System
• Multiple-Antenna System
• Mix of Dimensions
• Multi-carrier Multiple-Antenna System
• Ultra-wideband with Multiple-Antenna System
  (UWB-MEA) NEW
• All Dimensions
• Cognitive Radio NEW

7
Sub-sampled analytic UWB Radio (Mike Chen)

• Goal An ultra low-cost UWB radio
• Subsampling to remove local oscillator and mixer
  and reduce requirements on A/D.
• Analytical signal processing to ease timing
  recovery without oversampling and interpolation.

8
60 GHz Wireless System Design (David Sobel)
Clk

• Goal 1 Gigabit/Sec indoor picocellular network.
• Mostly-analog approach
• 2-CPFSK modulation embedded clock in I-channel
• Simple analog structures running at baud rate
• Directional antenna arrays for increased SNR,
  decreased multipath

9
Multiple-Antenna RF Front-End (Danijela Cabric)

• Goal a scalable multi-antenna front-end based on
  Intel 2.4 GHz radio connected to BEE for baseband
  processing.
• Features
• 12 bit precision _at_ 60 MHz
• 4 channel optical link each at 1Gbps
• Status design finished, fabrication is in
  progress

10
Energy/Delay-Efficient Circuits for MIMO(Dejan
Markovic)

• Goal investigate energy/delay-efficient
  implementation of multi-dimensional algorithms
• Implemented an SVD-based multiple-antenna channel
  decoupling algorithm on Simulink and BEE.
• Driver for Energy/Delay architecture optimization
  by using time-multiplexing and interleaving.

11
Understand Multi-Antenna Channels from a
Physics-based Perspective (Ada Poon)

• Goal incorporate antenna theory and physical
  propagation to derive more realistic limitations
  and insights into design.
• Propose to parameterize physical environment by
  number of clusters and cluster angle.
• Develop figure of merits on the optimal number of
  antenna elements given the physical environment
  and array size.
• Study the impact of scattering on performance
  tradeoffs.

12
Summary of Projects

• Individual Dimensions
• Ultra-wideband
• 60 GHz System
• Multiple-Antenna System
• Mix of Dimensions
• Multi-carrier Multiple-Antenna System
• Ultra-wideband with Multiple-Antenna System
  (UWB-MEA) NEW
• All Dimensions
• Cognitive Radio NEW

13
UWB with Multiple-Antennas?

• Design problems in UWB
• Multipath (delay spread gt 10 ns)
• Interference from other RF sources
• Low transmit power
• Multiple antennas provide
• Multipath reduction
• Interference cancellation
• Beamforming to increase power at receiver
• We have prior work on both, so why not see what
  we can do.

14
Physical Layer Multipath (Time/Space)

• Delay spread on each antenna element

RX
TX

• Delay spread on different angles of arrival

Shorter channel equalizers.
15
MAC Layer Interference (Freq/Space)

• Hidden terminal
• C attempts to transmit to B while A sends to B.
• C interferes with A.
• Multi-band orreceive beamforming
• Expose terminal
• C attempts to transmit to D while B sends to A.
• C is delayed.
• Multi-band ortransmit beamforming

A
B
C
16
Network Layer User Cooperation (User/Space)

• Route Communication through immediate nodes.
• Transceiver beamforming
• Create spatial channels to mimic wired
  cooperation.
• Save power at TX.
• Capture more power at RX.

Mesh Network
17
UWB-MEA Radio

• A multi-dimension (time, frequency, space and
  user) cross-layer radio
• Provide a learning experience on
• Multi-dimensional trade-offs on performance.
• Cooperation, Coordination and Co-design across
  different layers.
• Statues
• UWB spatial channel measurement and modeling.
• Multiple-antenna algorithm supporting transceiver
  beamforming and interference suppression.

18
UWB Spatial Channel Measurement and Modeling
(Jing Yang)

• Goal Provide real data for testing and
  verification, and characterize channel parameters
  for channel emulation.
• Features
• 4 rotational antennas
• 0-6 GHz BW
• 20 GSa/s per antenna
• Status
• Setting up the measurement and characterizing the
  wideband antennas.

19
A Two-Stage Low-DimensionalityMultiple-Antenna
Transceiver Design
TX Beamformer
RX Beamformer
Space-timeprocessing
SpatialChannelDecoupling
ClusteringChannel
InterferenceSuppression
Spatial Decimation
Spatial Interpolation
RelayingProcessing
ChannelEstimation
From RX
PhysicalEnvironmentLearning

• Channel estimation/decoupling
• SVD, QR,
• Physical environment learning
• Simple no. of clusters and cluster angle, no.
  of cooperative and non-cooperative users.
• Sophisticated cluster boundary, directions of
  users.

To TX
20
Summary of Projects

• Individual Dimensions
• Ultra-wideband
• 60 GHz System
• Multiple-Antenna System
• Mix of Dimensions
• Multi-carrier Multiple-Antenna System
• Ultra-wideband with Multiple-Antenna System
  (UWB-MEA) NEW
• All Dimensions
• Cognitive Radio NEW

21
Cognitive Radio

• According to FCC (Cognitive Radio Technologies
  Workshop 2003)
• A cognitive radio is a brainy radio that can
  sense and learn the environment.
• Paradigm shift from a passive receiver to a radio
  that can sense and learn, then find and adapt to
  the environment.
• FCC is going to release some spectra for this
  approach (like UWB it is an underlay strategy)

22
Cognitive Radios
Constraints
Dimension 2
Feasibleregion
Dimension 1

• Sensor finds the feasible region
• Optimizer selects the best waveform
• when, how long, frequency, bandwidth, array
  configuration.
• Reconfigurable baseband adapts to the optimal
  schemes.
• Mini-BEE adapts, UWB radio senses, and our
  missing piece is the brain!

23
Open Spectrum Flexible Radio

• In licensed spectrum, flexible radio can improve
  throughput but is not a necessity.
• If spectrum scarcity is due to inefficient use,
  flexible radio is a must for utilizing all the
  dimensions of spectrum use.
• So if spectrum scarcity is artificial, what about
  the OSI model?

Less flexible
More flexible
24
Another new project

• Scaling up the BEE

25
The question we then asked ourselves

• The success of the BEE has demonstrated the
  practicality of using a fully flexible
  architecture that is highly parallel.
• So
• If we have such a powerful general purpose
  computer why do we need a StrongArm for the
  simple I/O functions and GUI???

26
Taking BEE to the next level - TBEE

• General purpose computing model
• As easy to program (or even easier) than writing
  C/C for a conventional processor
• Demonstrate the capability of the technology to
  provide supercomputer performance
• A complete system solution, including
  computation, user interfaces, I/O, debugging and
  monitoring

27
A 3-Tiered Problem
User Interface

• Design environment
• Programming model
• Operating system (BORPH)
• Abstraction layer
• System architecture
• Clocking and interconnect

Operating System
Physical Hardware
28
TBEE

• Hardware Design Goals
• 1014 (100 TeraOperations) per second
• Able to exploit new FPGA technology (modular
  design)
• Robust against interconnect and component flaws
  (exploitation of homogeneous array redundancy)

29
Status

• UWB chip in final stages of design
• Gain at 60 GHz in 130 nm CMOS
• Board design for optical connectivity in final
  stages for multiple antenna transceiver
• Scaling up BEE
• And much, much more. (see the posters!)