Accurately tracking space debris and operational satellites is the foundation of the long-term sustainability of space operations. To improve upon some of the inherent limitations of ground radars, a constellation of satellites carrying optical sensors for the surveillance of the Low Earth Orbit (LEO) region is analysed. This analysis aims to understand the performance drivers of such a system in terms of constellation geometry and provide a general methodology for the preliminary design of the system. First, a method for decoupling the design of the optical payload and the constellation geometry while retaining statistically significant results is shown. Using the resulting estimate for the maximum observable distance, an approximate method for computing the coverage of the system is proposed. The expected daily and yearly variation of coverage depending on its own dynamics and the position of the Sun is analysed, showing that it has a small impact on the design process. The dependence of the coverage on constellation parameters such as altitude, inclination and distribution of satellites is investigated through parametric analysis, retrieving an estimate for the Pareto front of the system. Building upon the previous results, a random search method is shown to be effective in finding a design point lying on the Pareto front that is robust to both random satellite loss and deployment strategy. Finally, a reduced budget architecture is proposed to achieve acceptable performance while using only a few tens of satellites. The resulting work answers the problems of estimating and optimising the performance of a distributed system for space-based surveillance of the LEO region, a stepping stone for future cost–benefit analyses for the enhancement of space surveillance networks.