Você está na página 1de 17

Receive Diversity Revisited: Correlation, Coupling and Noise

Carlo P. Domizioli, Brian L. Hughes, Kevin G. Gard, and Gianluca Lazzi Department of Electrical & Computer Engineering North Carolina State University cpdomizi@ncsu.edu November 28, 2007
This material is based upon work supported by the National Science Foundation under grants CCF-0312686 and CCF-0515164

Motivation
Performance of receive diversity systems dependent on fading correlation between diversity branches independence is ideal In spatial diversity, receiver space constraints limit the antenna spacing, resulting in correlation between the multipath components and mutual coupling between the antennas Most research on mutual coupling focuses on fading (signal) correlation while assuming spatially white noise Performance metrics in communications depend on both the signal and noise; therefore we should also consider how closely spaced antennas affect noise correlation
2/14

Recent Work
Morris & Jensen 05: Realistic model for front-end amplifiers, compared matching networks optimized for power and noise Gans 06: Evaluated MIMO capacity for sky-noise limited and certain amplifier-noise limited scenarios p Goals of this study: Develop a receiver noise model that articulates the dominant sources of noise and includes effects of mutual coupling on both the signal and noise correlation g Evaluate performance of the optimal diversity receiver under a variety of different noise sources

3/14

Optimal Combiner for Correlated Noise p


Consider an M-branch diversity receiver in which both the fading and noise are spatially correlated:

r = hx + n
Fading path gain: (Correlated) AWGN:

x ~ Tx symbol; h, n C M
h ~ CN (0, h ) n ~ CN (0, n )

Determine the optimal (w.r.t. SNR) linear combiner Output SNR:

y = wHr

h H n1h, w eq iff w n1h w. eq.

Traditional MRC (w h) is suboptimal for correlated noise! Need a receiver noise model to determine specific form of n
4/14

Receiver Noise Model


Consider a typical post-detection diversity receiver:

...

...

...

Assume coupling in antennas only Each component contributes noise to the total output noise n Use noisy circuit theory to construct a noise model for each component, then calculate output noise correlation n
5/14

Antenna Array y
Model with a Thevenin equivalent network:

v = ZAi + u
ZA ~ antenna impedance matrix u ~ open-circuit (induced) voltage
Off-diagonal elements of ZA represent mutual coupling between antennas Terminal (observed) voltages related to incident field by array radiation pattern and impedance matrix:

6/14

Antenna Thermal Noise


Open-circuit voltage u contains a signal and noise component:

u = h o x + n o , h o ~ CN (0, h o )
Noise component no may represent induced thermal radiation or interference from other electronic devices For thermal noise in an isotropic environment (Twiss 55)

n o ~ CN (0, n o ), n o = 4kT0 B Re[Z A ]


kT0 4 10-21 W/H (at standard temp.); B ~ b d idth 10 21 W/Hz ( t t d d t ) bandwidth For antenna separations less than a few wavelengths ZA is non-diagonal noise i correlated! di l i is l t d!
7/14

Amplifier Noise p
Amplifiers typically represented by the Rothe-Dahlke (56) model:

va ~ CN (0,4kT0 Bra ) ia ~ CN (0,4kT0 Bg a )

ra ~ equivalent noise resistance q g a ~ equivalent noise conductance

This adequately models both thermal and shot noise Important amplifier metric is the noise figure NF:

SNR out = SNR in NF (in dB)


NF function of noise parameters {ra ,ga ,zcor} and source impedance
8/14

Downstream Noise
Downstream components consist of filters, mixers, amplifiers and other noisy circuits a detailed model would be too complicated Alternative assume each component performs a linear operation on the complex baseband signals and g p g generates AWGN Can reference total downstream noise to the amplifier output, model with a Thevenin equivalent load:

vd ~ CN (0,4kT0 B d ) Br

9/14

Matching Network g

Noise figure of amplifiers minimized iff Z in = z opt I (multiport match) Practical, suboptimum solution: Use M two-port MNs (self match) two port
10/14

Numerical Example p
Performance metric Diversity gain at 1% outage probability vs. antenna separation d Incident electric field 32 vertically-polarized plane waves: Angles-of-arrival uniformly spaced in azimuth from 0 to 2 g y p i.i.d. phases uniformly distributed on [0,2] Antenna array Two half-wavelength (/2) dipoles with radius 10-3. Impedance matrix and radiation pattern evaluated by NEC Amplifier Maxim 2642 LNA, NFmin=1.04 dB Downstream noise Assume components have a composite noise figure of 10 dB at a source impedance of 50 :

rd = rs (10

NF 10

1) = 450
11/14

Matching Network Performance g


13 12

Diversi Gain [d ity dB]

11 10 9 8 7 i.i.d. Fading & Noise Multiport Match Self Match 0 0.2 02 0.4 0 0.6 06 0.8 08 1

d/

12/14

Individual Noise Sources (Self Match) ( )


13 12

Divers Gain [dB] sity [

11 10 9 8 7 i.i.d. Fading & Noise Antenna Noise A t N i Amplifier Noise Downstream Noise 0 0.2 02 0.4 04 0.6 06 0.8 08 1

d/
13/14

Conclusions
In a compact receive diversity array, both the signal and noise components of the diversity branches may be correlated y y Traditional MRC is suboptimal for correlated noise Different noise sources can impact performance in profoundly different ways Accurately representing the dominant noise sources is critical to predicting performance in any multiple antenna receiver multiple-antenna

14/14

QUESTIONS?

References
1. M. L. Morris and M. A. Jensen, Improved network analysis of coupled antenna diversity performance, IEEE Trans. Wireless l d t di it f T Wi l Commun., vol. 4, pp. 1928-1934, Jul. 2005. 2. M. J. Gans, Channel capacity between antenna arrays Part I: Sky , p y y y noise dominates, IEEE Trans. Commun., vol. 54, pp. 1586-1592, Sep. 2006. 3. M. J. Gans, Channel capacity between antenna arrays Part II: Amplifier noise dominates, IEEE Trans. Commun., vol. 54, pp. 1983-1992, Nov. 2006. , 4. R. Q. Twiss, Nyquists and Thevenins theorems generalized for non-reciprocal linear networks, J. Applied Physics, vol. 26, pp. 599602, May 1955.

References (cont.) ( )
5. H. Rothe and W. Dahlke, Theory of noisy fourpoles, Proc. IRE, vol. 44, pp. 811-818, Jun. 1956 l 44 811 818 J 1956.

Você também pode gostar