Detection of TR UWB signal in the presence of Narrowband - - PowerPoint PPT Presentation

Detection of TR UWB signal in the presence of Narrowband Interference

Yohannes Alemseged and Klaus Witrisal yohannes@sbox.tugraz.at. Signal Processing and Speech Communication Laboratory spsc.inw.tugraz.at Graz University of Technology, Austria

UWB4SN workshop Nov. 4, 2005, Lausanne, Switzerland Detection of TR UWB signal in the presence of Narrowband Interference – p.1/23

Outline

• Introduction
• System Model: UWB signal, NBI signal, AcR front end
• Data model
• Detection Schemes
• Simulation Results
• Conclusion and further outlook

Detection of TR UWB signal in the presence of Narrowband Interference – p.2/23

Introduction

• Tx System:
• Transmitted Refference(TR) Differential UWB
• Receiver:
• TR AutoCorrelation Receiver (AcR)
• Interferor:
• IEEE 802.11a WLAN Service

Detection of TR UWB signal in the presence of Narrowband Interference – p.3/23

Introduction

• Why Interference could be adversary for UWB?
• unregulated spectrum
• should operate under -41.3dBm/MHz
• Why AcR is a choosen as a victim receiver?
• Transmitted Refference
• Analog front end / Nonlinearity
• Why IEEE 802.11a WLAN is choosen as interferor?
• Common spectrum
• Deployment scenario

Detection of TR UWB signal in the presence of Narrowband Interference – p.4/23

System Model

Figure 1: System Model

a

x x x Z Ts Dj Ts Ts Dj Integrate& WNcr W2 W1

s(t) β(t) ˆ r(t)

damp

Detection of TR UWB signal in the presence of Narrowband Interference – p.5/23

System Model: UWB signal

• UWB signal

• For differential TR scheme,

,

• Received Signal:

is AcR front end filtering effect (

)

Detection of TR UWB signal in the presence of Narrowband Interference – p.6/23

System Model: NBI Model

• NBI, IEEE802.11a OFDM passband signal

fc + Wβ

2

fc − Wβ

2

Pβ

Figure 2: Model of the narrowband interference

8 < :

• therwise

Detection of TR UWB signal in the presence of Narrowband Interference – p.7/23

System Model: Received Signal

• Correlator Output

Z

• Expanding further the correlator output equations;

Z

Z

Z

Z

Z

Detection of TR UWB signal in the presence of Narrowband Interference – p.8/23

System Model: Correlators Output

• NBI by NBI product term in the data sample

• First term can be neglected for

• ✱

is in GHz range and

is about

for indoor propagation

Detection of TR UWB signal in the presence of Narrowband Interference – p.9/23

Data Model

• Fixed phase sinusoid

modulated by a Short time autocorrelation function of the baseband NBI signal

• ✴

• Samples will be well correlated if

• For IEEE 802.11a WLAN

,

• NBI term can thus be written as;

• ✴

c is a sampled vector of

Detection of TR UWB signal in the presence of Narrowband Interference – p.10/23

Data Model

• Where ISI is avoided (burst /LDC), a data model can be

written as;

• ✰

• ✴

• ✁

is due to the "data by data" term, and

is a bias term due to interference among pulses

• In the absence of NBI and noise, detection could be

accomplished by;

• ✏

Detection of TR UWB signal in the presence of Narrowband Interference – p.11/23

NBI Mitigation: Least Square Solution

• Sampled output for

ymbols

• Find an estimate of

that minimizes the following norm

Setting the derivative of the cost function to zero

Implementation requires pilot symbols

Detection of TR UWB signal in the presence of Narrowband Interference – p.12/23

NBI Mitigation: Detection based on SVD

• can be factored as;
• ☎

Where

• ☎

is

x

data matrix

is

x

unitary matrix and

is

x

unitary matrix

where

• Correlation of the code vector

and the column containing the data from unitary matrix

is used for separation

Detection of TR UWB signal in the presence of Narrowband Interference – p.13/23

NBI Mitigation: Adaptive Constant Modulus Algorithm

• Constant amplitude
• ✁

• f binary antipodal

symbols

are transmitted

• The adaptation equation can be written as;

• Detection of TR UWB signal in the presence of Narrowband Interference – p.14/23

NBI Mitigation: Minimizing-Rank of Data Matrix

• The nearest minimum rank Matrix
• will be;
• ✲

• will be reconstructed by 0 substition or nulling the

smaller singular values

Detection of TR UWB signal in the presence of Narrowband Interference – p.15/23

NBI Mitigation:MMSE solution

• Detection Error

• ☎
• ✁

• ☎
• ✆

• ☎

• ✴

• Optimum combiner that minimizes

assuming

and noise is uncorrelated with the NBI and UWB signal, will be;

• ✝

• ✴

• Computing

requires knowledge of the model parameters

Detection of TR UWB signal in the presence of Narrowband Interference – p.16/23

Simulation Results:Parameters

NBI:

• signaling: OFDM
• carrier frequency:

5.125GHz

• no of carriers: 48
• baseband: QPSK

UWB:

• Tr: Differential-TR
• No of pulses/sym: 9
• Symbol Period: 100ns
• Delay: 0.2ns

1 2 3 4 5 6 7 8 −0.5 −0.4 −0.3 −0.2 −0.1 0.1 0.2 0.3 0.4 0.5 Sampled versions of the UWB, NBI and Noice at the receiver front end NBI Noise UWB

Figure 3: UWB (LDC), NBI and Noise at the front end

Detection of TR UWB signal in the presence of Narrowband Interference – p.17/23

Simulation Results:SVD Decomposition (low SIR)

20 40 60 80 100 120 140 −200 −100 Bias SIR=−20dB , Threshold BER=0.48, SVD BER=0.047 20 40 60 80 100 120 140 −100 100 20 40 60 80 100 120 140 −20 20 20 40 60 80 100 120 140 −20 20 Data

Figure 4: Sampled output of the correlators (sym-

bol axis)

1 2 3 4 5 6 7 8 −500 500 bias SIR=−20dB , Threshold BER=0.48, SVD BER=0.047 1 2 3 4 5 6 7 8 −200 200 1 2 3 4 5 6 7 8 −50 50 1 2 3 4 5 6 7 8 −5 5 data

Figure 5: sampled output of the correlators (cor-

rlators axis)

Detection of TR UWB signal in the presence of Narrowband Interference – p.18/23

Simulation Results:SVD Decomposition (high SIR)

20 40 60 80 100 120 140 −10 10 Data SIR=10, Threshold BER=0, SVD BER=0 20 40 60 80 100 120 140 0.2 0.4 Bias 20 40 60 80 100 120 140 −0.05 0.05 20 40 60 80 100 120 140 −0.05 0.05

Figure 6: Sampled output of the correlators (sym-

bol axis)

1 2 3 4 5 6 7 8 −2 2 Data SIR=10, Threshold BER=0, SVD BER=0 1 2 3 4 5 6 7 8 −2 2 Bias 1 2 3 4 5 6 7 8 −0.5 0.5 1 2 3 4 5 6 7 8 −0.5 0.5

Figure 7: sampled output of the correlators (cor-

relators axis)

Detection of TR UWB signal in the presence of Narrowband Interference – p.19/23

Simulation Results:SVD, LS BER

−30 −20 −10 10 20 30 10

−5

10

−4

10

−3

10

−2

10

−1

10 signal−to−interference ratio (SIR) [dB] bit error rate (BER) BER curves for Threshold detector and SVD detector Threshold detector SVD

Figure 8: BER plots for conventional threshold de-

tector vs SVD

−30 −25 −20 −15 −10 −5 10

−6

10

−5

10

−4

10

−3

10

−2

10

−1

10 signal−to−interference ratio (SIR) [dB] bit error rate (BER) BER curves for LS with training symbols, CMA (with codevector), Threshold detector threshold detector LS 288/288 LS 32/288 adaptive CMA

Figure 9: BER plots for LS solution and Adaptive

CMA (initialization with code vector)

Detection of TR UWB signal in the presence of Narrowband Interference – p.20/23

Simulation Results:CMA BER

−30 −25 −20 −15 −10 −5 5 10 10

−4

10

−3

10

−2

10

−1

10 signal−to−interference ratio (SIR) [dB] bit error rate (BER) SNR=35dB Threshold Detector LS with training sq CMA−ReducedRank CMA

Figure 10:

BER plots for 35dB SNR and SIR axis(model parameters used for initialization)

10 15 20 25 30 35 40 45 50 10

−6

10

−5

10

−4

10

−3

10

−2

10

−1

10 signal−to−noice ratio (SNR) [dB] bit error rate (BER) SIR=−10dB, Initialization vector LS soln at SNR=35dB Threshold Detector LS LS−TrSyb AdptCMA−MinRank AdptCMA

Figure 11: BER plots for -10dB SIR and SNR axis

Detection of TR UWB signal in the presence of Narrowband Interference – p.21/23

Conclusion

• We presented the analysis of the narrowband interference for

frame differential TR-UWB system and LDC scheme.

• A data model has been derived for LDC schemes which could

provide us a better understanding and insight for possible interference cancellation.

• In the presence of NBI the conventional/threshold detector

performs poorely.

• Detection based on SVD of the data matrix is found promising for

strong Interference (as compared to the conventional detector), however it gets unstable for higher SIR

Detection of TR UWB signal in the presence of Narrowband Interference – p.22/23

Conclusion

• Detection based on LS solution showed much more improved

performance than the conventional detector

• Detection based on adaptive CMA is shown to provide comparable

performance as that of the LS (no pilot symbols are used). During simulation it was observed that initialization vector is highly critical for the performance.

• This fact is attributed to the existence of multiple local minimas and

the CMA fails to attain a global minimum point.

• Further improvement of the mitigation schemes performance might

need tuning of the model to include nonlinear terms

Detection of TR UWB signal in the presence of Narrowband Interference – p.23/23