Distributed computing has become essential for processing large-scale computational tasks across interconnected devices and networks. However, its efficiency is often limited by communication costs, computational resources, and network connectivity. This talk explores how exploiting the mathematical structure of computational tasks can help overcome these limitations.
We begin with an information-theoretic perspective on distributed function computation, focusing on how the structure of Boolean functions and data placement influence communication requirements. We then extend the discussion to more general, nonlinearly separable functions, showing how sparse tensor factorization can be used to develop communication-efficient distributed computing schemes. Building on these theoretical foundations, we discuss recent developments in coded computation and their potential applications in resource-constrained networks.
Finally, we turn to emerging challenges in satellite and unmanned aerial vehicle (UAV) networks, where limited onboard resources, dynamic connectivity, and communication constraints motivate new approaches to distributed computation. We highlight ongoing research directions toward efficient and resilient computing across interconnected satellites.
Overall, the talk aims to bridge fundamental principles of distributed computation with the design of future networked computing systems, from terrestrial networks to space.