The basis of energy balancing method is regarding that the amplitude of azimuth spectrum is generally the same as the pattern of antenna [5]. Thus, the energy center of azimuth spectrum can represent the estimation of Doppler centroid.

The energy balancing method is proceeded by cyclic iteration. First of all, the initial value of Doppler centroid is set, and the Doppler centroid is calculated by iterative solving the energy of the both sides of the Doppler centroid. The principle of the energy balancing method is shown in Fig. 3.3. Specific processing steps are as follows.

(a) After the range compression of the raw data, transform the data into azimuth frequency domain;

(b) Set the initial Doppler centroid f^{0} by using the system parameters and the INS

data;

Fig. 3.3 Principle of the energy balancing method

(c) Calculate the energy of both sides E_{1} and E_{2}, and calculate the normalized energy difference AE = j^{1} +E^{2};

(d) Solve the relation coefficient s = ;

^{f}ae=0

(e) Set f+^{1} = fc + Af_{c}, where l is an integer. Repeat the cycle, until AE = 0. Then, the Doppler centroid estimation is f_{dc} solved.

Energy balancing method can provide an accurate Doppler centroid estimation in the homogeneous scene. However, if the scene is non-homogeneous, especially if prominent point targets exist in the scene, energy center cannot correctly replace Doppler centroid, and thus the energy balancing method is ineffective. One way to suppress the impact of the prominent point targets by using the average of multiple range gates. However, this method cannot eliminate the prominent point target, and the Doppler centroid relationship among different range gates is neglected. In [6], an improved energy balancing method is proposed by setting a threshold. This method can only eliminate targets which have a high energy peak than the threshold, and the continuity of the azimuth spectrum is destroyed.