Synthesis of wideband imaging beamformers using evolutionary programming

This paper presents spatial arrays of antenna elements deployed over a plane region are often used to image sources of reflected power in radar applications. The array of antennas samples the reflected field at a fixed set of locations. The received data are then processed using signal processing techniques such as beamforming to produce an image characterizing the distribution of the objects in the field of view of the active array imaging system. For some applications, as in through-wall microwave imaging, the scene to be imaged is often illuminated with wideband signals to achieve good range resolution. For such imaging problems, the concept of coarrays may be used to facilitate array signal processing issues in a simplified context [1, 2], resulting in a Point Spread Function, PSF(thetas, phi), which is not only a function of weights at the coarray elements but also a function of the wideband signal spectrum, S(omega). For high resolution imaging, it is desirable to synthesize the weights and S(omega), to achieve a desired PSF distribution, for example one with certain beam shape and/or with tapered side lobe levels. This could be potentially a challenging non-linear optimization problem, consisting of a very large number of optimization parameters, and is different from the standard narrow-band fixed antenna array problem, which maybe solved by traditional array synthesis techniques. Evolutionary algorithms such as genetic algorithms (GAs), evolutionary programming (EP) and particle swarm optimization (PSO) are well suited for this class of problems.