As the International Telecommunications Union (ITU) IMT-2030 have just begun invitation to participate in the global study surrounding the use-cases for 6G, the technical Key Performance Indicator (KPIs) which translate from the potential use-cases have not been explicitly reported. However, with the innately higher speeds, lower latency and improved reliability planned for 6G cellular technology a variety of new use-cases and applications will become possible, including sensitive time-critical environments such as Machine to Machine (M2M), Medical and Military. Supporting the advancements in technology surrounding human-centric, environmental-centric, machine-centric and sustainability-centric applications, 6G is the solution for increased area traffic capacity, energy efficiency, network density, latency, mobility, spectral efficiency and data rates. Light communications offer a range of innate advantages, including wide, and virtually unlimited, bandwidth and solid state hardware. As well as this, the plan for 6G 100GHz communications are generally Line Of Sight (LOS) which is ideal for Optical Wireless Communications (OWC). In contrast to traditional Radio Frequency (RF) communication systems where complex envelopes mixed onto a carrier signal for transmission are required, OWC systems use real signals which operate at base-band. As such, many works have been completed in developing signal standards appropriate for light communications. This thesis contributes to the advancement of such OWC signal standards with respect to spectral efficiency, channel estimation and equalization, Multiple In Multiple Out (MIMO) architectures and physical implementation. The novel developments listed are cultivated with the acknowledgement of pre-existing, as well as complimentary traditional RF, technologies. The focal point of the work presented here is the optimisation of techniques for signal transmission and real-time Field Programmable Gate Array (FPGA) transceiver development. The first novel contribution is a technique to increase the spectral efficiency of OWC signals. By taking advantage of an existing characteristic of most OWC signals, the spectrum efficiency is increased by the removal of redundant information which can be reconstructed in the Receiver (Rx) to obtain the original data set. By using the novel processing algorithm cultivated from the innate Hermitian Symmetry behavioural characteristics within OFDM-based optical signals, the spectral efficiency is doubled. The second contribution is the development of a low-latency FPGA OWC transceiver. As most OWC techniques are low-complexity, the resources available on-board can be utilized carefully to achieve lower latency systems. This is a major requirement of developing 5th Generation (5G) and 6th Generation (6G) technologies which include the use-case of critical systems to their standards. Measured results consist of a sub-400ns Transmitter (Tx) and a sub-1ms Rx. The next contribution is the optimization of a single iteration channel estimation and equalization technique using the Zadoff-Chu sequence for OWC which has multipurpose potential. It is shown experimentally that this sequence can be used for time synchronisation, physical encryption and channel distortion correction. This technique exhibits performances similar to more complex estimation methods such as Recursive Least Squares (RLS) and Least Mean Squares (LMS) which converge after multiple loops. The next contribution involves the minimisation of Digital Pre-Distortion (DPD) computational demands in MIMO systems. Although linear Power Amplifiers (PAs) are often used in optical front-ends, as research and the demands for distance increase, so will the output required from the PA modules such as within Millimeter Wave (mmWave) RF chains. By scaling the signals to a reference point, only one DPD path is required for a multi-path system, as opposed to an individual DPD block for each path. This novel technique is experimentally validated on a 4-PA array with DPD trained on a singular PA, minimising the ACPR to acceptable transmission levels within MIMO architecture. The final contribution involves the minimisation of the effective constellation size of a signal to increase the overall average power, leveraging the best dynamic ranges for any given channel scenario.