Adaptive Programmable Integrated Microwave Photonic Signal Processor （Invited）

With the explosive growth of next-generation information technologies， such as ultra-high-speed 5G/6G wireless communication networks， ultra-wideband radar systems， and quantum information processing， the requirements for the generation， transmission， and manipulation of high-frequency microwave signals have reached unprecedented levels. Facing the escalating challenges of bandwidth， speed， and energy efficiency， Integrated Microwave Photonics （IMWP） has emerged as a transformative solution. By leveraging the ultrawideband characteristics of photons propagating within integrated circuits， IMWP achieves high-precision modulation， distribution， and processing of microwave signals through interference and multiplexing across diverse optical paths， thereby surmounting the performance bottlenecks inherent in traditional electronic systems. Furthermore， on-chip programmable photonic-microwave signal processors provide a robust hardware foundation for next-generation high-performance microwave systems by reconfiguring their internal architectures to achieve multi-functionality and flexible configuration. Although various architectures based on interferometer arrays， such as finite impulse response filters with delay-line arms and cascaded ring resonators， have been demonstrated， a critical bottleneck hindering their engineering transition is the difficulty in balancing architectural reconfigurability with system scalability in complex application scenarios. Conventional designs often suffer from fixed interconnection topologies and a lack of directness in functional implementation， which constrains their practical deployment in diverse and dynamic environments.Addressing these limitations， this work proposes and experimentally demonstrates an adaptive programmable integrated photonic signal processor based on a hexagonal waveguide mesh topology. The processor utilizes Mach-Zehnder Interferometers （MZIs） as fundamental tuning units， featuring an optimized physical-layer design to ensure high-fidelity functional mapping across the network. The device was fabricated on a 220 nm Silicon-on-Insulator （SOI） platform， with a compact footprint of approximately 3 mm×1 mm， integrating 32 tunable MZIs， 32 thermo-optic phase shifters， and 24 optical I/O ports. Comprehensive characterization of the MZI test structures reveals excellent performance， including extinction ratios exceeding 20 dB and a half-wave heating power of 4.41 mW. A central innovation of this study is the implementation of a self-adaptive control framework driven by the Crested Porcupine Optimizer （CPO） algorithm. The operational logic of the CPO-based configuration is divided into three distinct phases： initialization， closed-loop evaluation， and strategy evolution. By generating a random population within the predefined voltage range and iteratively calculating fitness values based on real-time experimental spectral feedback， the system autonomously converges to the optimal voltage configuration. During the evolution process， the algorithm dynamically toggles between exploration and exploitation modes using four distinct evolutionary strategies， ensuring both global search coverage and local refinement precision. This self-adaptive framework enables the processor to achieve immediate operational readiness with minimal control complexity， establishing a highly stable hardware-software co-design for universal signal processing.Experimentally， the fabricated processor was configured to demonstrate two representative microwave photonic functionalities： topological filtering and frequency measurement. In the topological filtering experiments， the mesh’s high degree of reconfigurability was leveraged to synthesize diverse spectral responses， including both Finite Impulse Response （FIR） and Infinite Impulse Response （IIR） filters. By precisely controlling the power splitting ratios and phase shifts within the hexagonal units， we successfully realized Asymmetric MZIs （AMZIs） with tunable extinction ratios exceeding 25 dB and Free Spectral Ranges （FSR） ranging from 0.17 nm to 0.64 nm. Furthermore， Optical Ring Resonators （ORRs） with a 6-BUL （Basic Unit Length） cavity and two-stage Coupled Resonator Optical Waveguides （CROWs） were implemented， demonstrating the mesh's capability to synthesize complex recursive structures with flattened transmission spectra. For frequency measurement， a broadband frequency identification system was constructed using an Amplitude Comparison Function （ACF） based on a Carrier-Suppressed Single-Sideband （CS-SSB） modulation scheme. The system utilized the mesh's ability to switch between multi-stage frequency discrimination channels with different FSRs. By synergistically combining the results from a wideband coarse-measurement channel and a narrowband high-precision channel， the system achieved a broad frequency identification range from 0.01 GHz to 30 GHz with a significantly improved Root Mean Square Error （RMSE） of 131 MHz. This collaborative measurement scheme effectively eliminates the periodic ambiguity typical of narrowband discriminators while maintaining high resolution across the entire spectrum.In summary， this work provides a viable and scalable pathway for realizing high-performance， low-complexity integrated microwave photonic systems. The hexagonal mesh architecture demonstrates exceptional functional adaptability and structural scalability， marking a paradigm shift from Application-specific Integrated Circuits （ASPICs） to software-defined universal photonic processors. Beyond the filtering and frequency measurement demonstrations， the architecture serves as a general-purpose hardware platform capable of supporting diverse tasks， including tunable delay lines for wideband beamforming， arbitrary waveform generation， and reconfigurable optoelectronic oscillators. Theoretical analysis and experimental validation suggest that this self-adaptive framework can be extended to larger-scale networks， providing a clear evolutionary path for multi-functional， adaptive measurement systems in complex electromagnetic environments. The successful implementation of these functionalities on a single programmable chip paves the way for the future deployment of integrated photonics in advanced radar systems， 5G/6G communication infrastructures， and high-dimensional optical signal manipulation.