The estimation of direction of arrival (DOA) is widely used in military and commercial communication applications. In military applications, for example, the estimated DOA is used for antijamming design of the global positioning system (GPS). The DOA estimation is also used in mobile communication systems to expand the coverage area or to increase the channel capacity, where the base station forms a beam to the direction of the user terminal in order to avoid interference to other terminals.
For DOA estimation, an antenna array is essential. By using the phase difference between the output signals of antennas, we can derive the DOA. In practice, however, mutual coupling between the elements of an antenna array change the beam pattern of each element and degrade the performance of DOA estimation. Especially, the estimation of signal parameters via rotational invariance techniques (ESPRIT) algorithm is seriously affected by the coupling.
In ESPRIT, two or more subarrays are used, and the elements of each subarray should have identical beam patterns. The different beam patterns impair the translational invariance structure between the subarrays and degrade the accuracy of DOA estimation.
The coupling effect depends on the position of the elements in the array, which results in different beam pattern of each element. However, it is a noteworthy fact that the mid-elements undergo relatively same coupling effect. And the same coupling does not affect DOA estimation.
In this paper, we use the mid-array ESPRIT to compensate the mutual coupling. And we estimate the mutual coupling matrix gradually in an adaptive manner. If we assume N-elements, the proposed method proceeds as follows: First, we apply the M mid-array ESPRIT to estimate the DOA where the M elements in the mid-array have similar beam patterns. The other (N-M)/2 elements are located at the each side of the mid-array. Second, we estimate the coupling coefficient and compensate the mutual coupling. Third, we apply the full array ESPRIT to estimate the DOA. By performing the second and third steps iteratively, we can improve the reliability of coupling compensation and correspondingly the accuracy of DOA estimate. Also, we can get more stable DOA estimate by using gradual update of the estimate.