The introductions of Digital Beam Forming (DBF), original signal exploitation and waveform multiplexing techniques have led to the design of novel radar concepts. Passive Coherent Locator (PCL) and Multiple-Input Multiple-Output (MIMO) sensors are two examples of innovative approaches. Beside the inherent benefits of this class of sensors, related to the use of emitters of opportunity and the waveform diversity for the PCL and the MIMO cases respectively, an additional gain is achieved by using ad-hoc array signal processing techniques. The well-known advantages provided by the antenna arrays, or phased arrays, are, among other, the possibility of synthesising arbitrarily steered beams, especially in the case of DBF, the capability of shaping the array pattern sidelobes region and the opportunity of placing nulls towards specific directions. The last feature allows suppressing, or mitigating, the echoes of unwanted signals (clutter, jammer, interferences) that are present inside the radar scenario and that can degrade the detection performance of the system. With DBF, the interferences suppression can adaptively be done in real time. This thesis has focused on the development of novel array processing algorithms, with emphasis on antenna pattern optimisation. Also, attention was paid to resolution enhancement and clutter suppression techniques exploiting the available radar architectures. A PCL and a MIMO array system have been investigated in order to validate the proposed techniques. Specifically, the first part of the thesis deals with the development of the array signal processing chain of a PCL system based on a circular array. The performance of digital beamforming algorithms highly depends on the behavior of the antenna section of the array; indeed, an inaccurate characterisation of the element channels and their mutual influence degrades the shape of the synthesised array pattern. A complete analysis of the circular array is presented in the text and a technique for the mutual coupling compensation, based on an optimisation approach, was retrieved. The effectiveness of the method was validated. Furthermore, a new pattern shaping technique for the sidelobes reduction was developed. Due to the poor angular resolution of the PCL array, the criteria for estimating the quality of the retrieved algorithm was represented by the achievable sidelobes suppression level for a given angular resolution. A comparison with the phase modes approach has proven the better performance of the proposed method. Concerning the validation of the PCL system, the entire signal processing chain of the sensor was developed and the related behavior was successfully assessed by the comparison with the ground truth data provided by an ADSB transponder. The second part of the thesis refers to coherent MIMO radars and to the benefifits they provide with respect to conventional array-based radars. The linear array confifiguration was taken as reference and a MIMO test board, the RADOCA system, has been realised and studied. By first recalling the peculiarity of such systems, i.e. the possibility of synthesising virtual array structures, the research activity has investigated the effect that illumination errors produce on the related pattern behavior. The analytical expression of the errors affected pattern was retrieved and the comparison with conventional linear array systems showed the higher degradation they produce on MIMO based arrays. The focus was then moved to the identification of calibration and array processing techniques, capable of exploiting both the MIMO radar principle and the additional degrees of freedom provided by the virtual channels. By referring to the RADOCA demonstrator, a calibration procedure was retrieved and validated thanks to real data acquisitions. Then, with the aim of enhancing the system performance, two objectives were reached: the extension of the maximum unambiguous Doppler interval, by means of a random selection of the radar transmitter sequence, and the identification of a novel 2D high resolution technique based on the MUSIC algorithm. The comparisons with traditional approaches highlighted the benefits of the proposed techniques.