The capacity of optical communications networks continues to grow unabated. Applications for streaming video, social networking and cloud computing, are driving exponential growth of the traffic carried over the world’s ICT networks, which has been sustained thus far through the proliferation of datacenters and efficient, effective use of existing optical fibre. To meet increasing capacity demands requires increasingly sophisticated modulation formats and spectral management to achieve effective use of the available spectrum provided by an optical fibre. Moreover, the technology developed for optical communications is finding broader application to other sectors such as data centres, 5&6 G wireless; lidar and radar. Ultimately, some essential signal processing functions must occur at speeds beyond purely electronic means even when accounting for anticipated technological development. The option is to perform signal processing in the optical domain. Optical signal processors are fundamentally analog and linear in nature. To provide high performance, an analogue processor must be well controlled in a way analogous to the numerous and sophisticated controllers employed by the process industry. Consequently, a further extension of control to deeper levels within the physical layer reaching the optical layer will be necessary. For example, current reconfigurable optical add-drop multiplexers are coloured and directional and the wavelength division multiplexing channel grid, transponders modulation format, and the routing are all fixed. Through optimization of the interface between the physical components, sensors, and processors elastic optical network technology can be achieved by employing colour-, direction-, contention-, grid-less, filter-, gap-less reconfigurable optical add-drop multiplexers, flexible channels centre frequencies and width, flexible sub-carriers in super-channels, flexible modulation formats and forward error control coding transponders, and impairment-aware wavelength routing and spectral assignment. The aim of this thesis is to advance the state-of-the-art in photonic circuits and subsystems via proposing new architecture; study of the feasibility of photonic integration and, proof of concept implementations using available resources. The goal is to introduce new architectural concepts that make effective use of physical components and/or optical processors with reduced energy consumption, reduced footprint and offer speed beyond all-electronic implementations. The thesis presents four case studies based on one or more published papers and supplementary material that advance the goal of the thesis. The first study presents a coherent electro-optic circuit architecture that generates N spatially distinct phase-correlated harmonically related carriers using a generalized Mach-Zehnder Interferometer with its N×1 combiner replaced by an N×N optical Discrete Fourier Transform. The architecture subsumes all Mach-Zehnder Interferometer-based architectures in the prior art given an appropriate selection of output port(s) and dimension N, although the principal application envisaged is phase-correlated subcarrier generation for next-generation optical transmission systems. The theoretical prediction is then verified experimentally using laboratory available photonic integrated circuit fabricated for other applications. Later on, a novel extension of the circuit architecture is introduced by replacing the optical Discrete Fourier Transform network using the combination of a properly chosen phase shifter and single MMI coupler. The second study proposes two novel architectures for an on-chip ultra-high-resolution panoramic spectrometer and presents their design, analysis, integration feasibility, and verification by simulation. The target application is to monitor the power of a wavelength division multiplexed signals in both fixed and flex grid over entire C-band with minimum scan time and better than 1 GHz frequency accuracy. The two architectures combine in synchrony a scanning comb filter stage and channelized coarse filter. The fine filtering is obtained using a ring resonator while the coarse filtering is obtained using an arrayed waveguide grating with appropriate configuration. The fully coherent first architecture is optimised for compactness but relies on a repeatable fabrication processes to match the optical path lengths between a Mach-Zehnder interferometer and a multiple input arrayed waveguide grating. The second architecture is less compact than the first but is robust to fabrication tolerances as it does not require the path length matching. The third study proposes a new circuit architecture for single sideband modulation or frequency conversion which employs a cascade Mach-Zehnder modulator architecture departing from the orthodox dual parallel solution. The theoretical analysis shows that the circuit has 3-dB optical and 3-dB electrical advantage over the orthodox solution. The 3-dB electrical advantage increases the linear operating range of Mach-Zehnder modulator before RF amplifier saturation. An experimental verification of the proposed architecture is provided using an available photonic integrated circuit. The proposed circuit can also perform complex modulation. An alternative implementation based on polarization modulators is also described. The fourth study presents the theoretical modelling of a photonic generation of broadband radio frequency phase shifter. The proposed phase shifter can generate any phase without bound: the complex transmission of the phase shifter follows a trajectory that rotates on a unit circle and may encircle the origin any number of times in either direction, which has great utility in the tuning of RF-photonic systems. The proposed concept is then verified experimentally using off the shelf low frequency electronic components.