
- •Contents
- •Preface
- •1.1 Smart Antenna Architecture
- •1.2 Overview of This Book
- •1.3 Notations
- •2.1 Single Transmit Antenna
- •2.1.1 Directivity and Gain
- •2.1.2 Radiation Pattern
- •2.1.3 Equivalent Resonant Circuits and Bandwidth
- •2.2 Single Receive Antenna
- •2.3 Antenna Array
- •2.4 Conclusion
- •Reference
- •3.1 Introduction
- •3.2 Data Model
- •3.2.1 Uniform Linear Array (ULA)
- •3.3 Centro-Symmetric Sensor Arrays
- •3.3.1 Uniform Linear Array
- •3.3.2 Uniform Rectangular Array (URA)
- •3.3.3 Covariance Matrices
- •3.4 Beamforming Techniques
- •3.4.1 Conventional Beamformer
- •3.4.2 Capon’s Beamformer
- •3.4.3 Linear Prediction
- •3.5 Maximum Likelihood Techniques
- •3.6 Subspace-Based Techniques
- •3.6.1 Concept of Subspaces
- •3.6.2 MUSIC
- •3.6.3 Minimum Norm
- •3.6.4 ESPRIT
- •3.7 Conclusion
- •References
- •4.1 Introduction
- •4.2 Preprocessing Schemes
- •4.2.2 Spatial Smoothing
- •4.3 Model Order Estimators
- •4.3.1 Classical Technique
- •4.3.2 Minimum Descriptive Length Criterion
- •4.3.3 Akaike Information Theoretic Criterion
- •4.4 Conclusion
- •References
- •5.1 Introduction
- •5.2 Basic Principle
- •5.2.1 Signal and Data Model
- •5.2.2 Signal Subspace Estimation
- •5.2.3 Estimation of the Subspace Rotating Operator
- •5.3 Standard ESPRIT
- •5.3.1 Signal Subspace Estimation
- •5.3.2 Solution of Invariance Equation
- •5.3.3 Spatial Frequency and DOA Estimation
- •5.4 Real-Valued Transformation
- •5.5 Unitary ESPRIT in Element Space
- •5.6 Beamspace Transformation
- •5.6.1 DFT Beamspace Invariance Structure
- •5.6.2 DFT Beamspace in a Reduced Dimension
- •5.7 Unitary ESPRIT in DFT Beamspace
- •5.8 Conclusion
- •References
- •6.1 Introduction
- •6.2 Performance Analysis
- •6.2.1 Standard ESPRIT
- •6.3 Comparative Analysis
- •6.4 Discussions
- •6.5 Conclusion
- •References
- •7.1 Summary
- •7.2 Advanced Topics on DOA Estimations
- •References
- •Appendix
- •A.1 Kronecker Product
- •A.2 Special Vectors and Matrix Notations
- •A.3 FLOPS
- •List of Abbreviations
- •About the Authors
- •Index

92 Introduction to Direction-of-Arrival Estimation
|
|
= arg(φ |
|
) and θ |
|
|
|
|
λ |
|
|
|
μ |
|
|
|
= arcsin |
− |
|
|
μ , 1 |
≤ i ≤ d |
(5.24) |
||
|
|
|
|
πΔ |
||||||||
|
i |
|
i |
|
i |
|
2 |
ι |
|
|
A brief summary of standard ESPRIT computation steps is shown in Table 5.1 [1].
As seen from Table 5.1, the standard ESPRIT computations involve complex number operations. To avoid this, an improved ESPRIT employing only real-valued computations has been developed. It is described in Section 5.4.
5.4 Real-Valued Transformation
The standard ESPRIT is comprised of complex-valued computations throughout the algorithm. This makes the algorithm computationally expensive. However, it is observed that if centro-symmetric arrays are used, the signal subspace can be estimated by real-valued computations. Exploiting this property for centro-symmetric arrays leads to the unitary ESPRIT with only real-valued computations. Centro-symmetric arrays introduced in Chapter 2 thus are essential for implementing unitary ESPRIT. Unitary ESPRIT is formulated and based on the following theorem [5].
Table 5.1
Summary of Standard ESPRIT Algorithm
1. Signal Subspace Estimation: Compute Us
as the d dominant left singular vectors of X (square-root approach) or the d dominant eigenvectors of XXH (covariance approach)
2.Solution of the Invariance Equation: Solve the following equation for
J1Us Ψ ≈ J2Us
by means of least-squares or total least-squares techniques
3.DOA Estimation: Calculate the eigenvalues of the resulting complex-valued solution
Ψ = TΦT−1
and then extract the angular information via
μi |
= arg(φ i ), |
1≤ i ≤ d |
|||
|
|
|
|
λ |
|
θ |
|
= arcsin |
− |
|
μ |
|
2π |
||||
|
i |
|
|
i |

DOA Estimations with ESPRIT Algorithms |
93 |
|
|
Theorem 5.1
Let Qp and Qq denote the left Π-real matrices of p × p and q × q, respectively, defined in (3.16). Then given any p × q centro-Hermitian matrix G, Q −p1GQ q is a real-valued p × q matrix.
Consider the extended data matrix defined in (4.8). It is written here again for ready reference.
Z = [X Π M |
|
Π N ] |
(5.25) |
X |
It can be shown that Z is centro-Hermitian [6]. Also, similar to what was described in Section 4.2, the estimate R xxfb of the forward-back-
ward averaged covariance matrix is centro-Hermitian [6].
fb |
|
1 |
|
|
|
|
Π M ) |
|
|
|
|
|
|
||||
R xx |
= |
2 |
(R xx |
+ Π M R xx |
(5.26) |
|||
|
|
|
|
|
|
|
|
Now, the unitary transformation of the complex-valued data matrix to real-valued matrix can be obtained by applying Theorem 5.1 to the extended data matrix.
Γ(X ) = ϕ(Z) = Q MH [X Π M |
|
Π n ]Q 2N R M × 2N |
(5.27) |
X |
The transformation Γ(X) thus accomplishes forward-backward averaging by first extending the complex-valued data matrix X of size M × N to a complex-valued centro-Hermitian matrix Z of size M × 2N and then transforming Z into a real-valued matrix of the same size. As a result, we can obtain
|
fb |
|
|
H fb |
|
|
|
1 |
|
|
H |
H |
|
|
|
Π M Q M ) |
||
|
|
|
|
|
|
|
|
|
|
|||||||||
ϕ(R xx ) = Q M R xx Q M = |
2 |
(Q M R xx Q M |
+ Q M |
Π M R xx |
||||||||||||||
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
H |
|
|
|
|
|
|
H |
|
|
|
|
|
|||
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|||||
|
= |
2 |
(Q M R xx Q M |
+ Q M R xx Q M ) |
|
|
|
(5.28) |
||||||||||
|
|
|
|
|
|
M } |
|
|
|
|
|
|
|
|
|
|||
|
= Re |
Q H R |
xx |
Q |
R M × M |
|
|
|
|
|
||||||||
|
|
|
|
{ M |
|
|
|
|
|
|
|
|
|
|
As seen, the forward-backward averaging can be achieved automatically and implicitly by taking the real part of the transformed covariance