Fast Beamforming With Fault Tolerance in Massive Phased Arrays Using Intelligent Learning Control

Abstract

This paper presents a novel algorithm called the recursive intelligent optimizer (RIO) applied to a digital phase control in dynamically reconfigurable phased arrays with applications mainly in radar systems and communication antennas. The RIO belongs to the class of iterative learning control algorithms and operates in deterministic, stochastic, and population initialization modes which are valuable in fast dynamic array reconfiguration. The array factor of a centro-symmetric array is modeled by a linear transformation and mainly optimized for the sidelobe level (SLL), beam steering, and interference suppression. The RIO dynamic reconfiguration is achieved in less than 1.2 s for an array of 100 elements where initial phases are optimized with recursive learning adding up to 5.7 and 8.5 dB to the SLL of 100 and 200 elements arrays, respectively. Failed phase shifters are turned off and the array synthesis is reoptimized for new digital phases with a minimal performance loss. Indeed, the failure of 10% of elements producing a heavy loss of 5.9 dB is fully compensated to a minor loss of 0.3 dB in 441 ms.