WHYNET: UltraWideband Physical Layer

1
WHYNETUltraWideband Physical Layer
Progress Meeting July 23, 2004 Urbashi
MitraDepartment of Electrical EngineeringUnivers
ity of Southern CaliforniaLos Angeles,
CA ubli_at_usc.edu Thanks to Mr. Stefan Franz and
Mr. Majid Nemati
2
UWB Research Focus

• Testbed Development
• Interfaces with the existing radios
• Simulation platform
• Characterization of physical layer for use at
  higher layers
• Characterization of radios
• Interface between simulation platform and other
  layers in the testbed

3
Leveraging the UltRa Lab

• UltRa Lab currently is predominantly a
  measurement facility
• Goal is to move to communications and measurement
• Recent activity
• Antenna design
• Antenna characterization
• Propagation measurements
• Propagation Measurement warehouse

4
Timeline and Deliverables

• Year 1
• Assess models for simulation environment
• Simulation code modularized, models collected by
  end of Summer 2004
• Integration with ns-type simulator by end of 2004
• Design simple synchronization
• New Transmitted Reference symbol and frame
  synchronization scheme designed
• Theoretical analysis ongoing
• Procure UWB radios from XtremeSpectrum
• Characterize physical layer performance and
  capabilities of XtremeSpectrum radios

5
Freescale Radio Specifications

• Variable data rate
• 25, 50, 75, 100 Mbps
• 1, ½, ¾ coding rates
• 802.15.3 protocol
• DS-CDMA spreading
• BPSK modulation
• Less than 1 mW radiated over 3.1 10.6 GHz
  (200mW)
• 10m range

6
Wireless Evaluation Kit
7
Signals in UWB

• Transmitted signal has the form
• P(t) is the transmitted template,
• e.g., second derivative of a Gaussian pulse.
• NuNumber of users
• NsyNumber of Tr symb
• NsNumber of symb/data symb
• NfNumber of frames/symb

8
UWB Signaling and Channels
9
UWB Testbed Effort

• UWB physical layer simulation
• Link with network simulator (NS) environment
• C programming
• Different settings and parameter selection
• Simulate different typical channels
• (IEEE 802.15 four UWB channel models)
• Different parameter selections in transmitter
• Different receiver structures

10
Simulation System Diagram
11
Channel Models

• Received Signal
• Transmitted signal
• Channel effects (Multipath, differentiation, etc)
• Narrowband interference
• Filtering (channel, Tx and Tr)
• Channel Models
• CM1 Line-of-Sight (LOS), 0-4m channel
  measurement
• CM2 Non-LOS (NLOS), 0-4m channel measurement
• CM3 NLOS, 4-10m channel measurement
• CM4 Extreme NLOS multipath, 25 nsec rms delay

12
Basic parameters (adjustable)

• Nsy Number of data symbols/observed interval
• Nd Number of transmitted symb/data symb
• Tsam Template sampling time
• Nu Number of users
• Ts Symbol time
• Nf Number of frames/symb
• Nc Number of chips/symb
• Nh Time hopping seq. max value
• Nr Number of reference symbols
• Delta-samTime of RAKE fingers
• Alpha Fraction of transmitted energy spent for
  training and data symbols
• Np period of spreading sequence (DS case)
• Bcode Block code being used

13
Clustered Channels
Cramer, Scholtz, Win
in time of arrival versus angle of arrival space,
multipath occurs in clusters
14
Sparse/Clustered Channel models

• Sparse channels versus dense channels
• Accepted channel models based on clusters
• A pragmatic way to reduce complexity and improve
  performance (tradeoff between number of
  parameters to estimate and amount of data)

we get gains of up to 4 dB by exploiting
clustering
15
Transmitted Reference Systems

• Why transmitted reference?
• Waste of information rate
• Waste of transmission power
• Performance degradation due to noisy template
  signal
• Advantages
• No explicit channel estimation needed
• High energy capture possible
• Example
• Our optimized TR designs over 4 dB gains over
  existing systems

16
Optimized Transmitted Reference

• Significantly improved BER (almost 4dB)
• Approximate receivers perform close to true
  receivers
• Small degradation for high SNR

accurate (modest complexity) approximations of
performance have also been designed
17
Optimized Integration Interval

• Integrating over the entire delay spread can lead
  to 1-3 dB losses in performance

• An optimal integration interval exists
• Yields complexity reduction

18
UWB Timing Synchronization

• Idea Semiblind approach
• Jointly exploit known pilot symbols and unknown
  data
• Use less pilot symbols
• Take advantage of unknown data symbols
• Transmitted signal
• data symbols
• transmitted monocycle
• time-hopping code
• , frame-time and chip time

19
Symbol and Frame Synchronization

• Semiblind approach
• Jointly exploit known pilot symbols and unknown
  data
• Use less pilot symbols
• Take advantage of unknown data symbols

• time-hopping code misalignment
• symbol offset

20
UWB Timing Synchronization

• Received signal
• channel response
• delay offset in multiples of the frame time
• delay offset in multiples of the sampling time
• Joint estimate of channel and timing offset

unknown
21
UWB Timing Synchronization

• Maximize likelihood function for and
• Soft-estimates on the unknown data
• Recursive structure since is dependent on
• Backsubstitution and low SNR approximation
• Timing estimates

22
Synchronization Operation

• Maximum likelihood based methods
• Approximations to reduce complexity

function of channel estimate
23
Probability of Synchronization

• Comparsion with DD scheme that uses hard
  decisions on the unknown symbols
• Semiblind scheme shows significantly improved
  probability of acquisition for small number of
  training symbols

24
BER for ML and GLRT Receiver

• BER for ML and GLRT receiver approaches perfect
  timing case (PTK)
• GLRT outperforms ML by about 1dB
• Added cost is complexity