Distributed Beamforming using Software-Defined Radios : Frequency and Phase Synchronization

This thesis investigates the feasibility and performance of distributed beamforming with geographically dispersed software-defined radios, targeting the challenge of achieving coherent transmission and an effective signal combining at the receiver. The project was carried out at the Swedish Defense Research Agency (FOI). In contrast to centralized systems, the distributed approach relies on independently clocked nodes, each with its own local oscillator. While this enables scalability, fault tolerance and cost-effectiveness, it also introduces major synchronization challenges in time, frequency and phase. The research is driven by the questions of how distributed beamforming can be achieved and what information must be exchanged among nodes. The study also examines the impact of the wireless channel where a simulation framework is developed, combining theoretical modeling, algorithm design and simulation-based evaluation. Three primary synchronization phases are investigated: Frequency synchronization based on frequency estimation, an iterative phase alignment, and frequency offset tracking based on extended Kalman filter. These operate within a master-slave architecture coordinated via time-division multiple access. Three frequency synchronization methods are evaluated, all outperforming unsynchronized baselines. The best results under moderate frequency offsets are achieved by combining two methods. A distributed frequency consensus method also shows potential, though at the cost of increased overhead. On the other hand, phase synchronization is achieved through an iterative one-bit feedback method, which scales efficiently and converges reliably. The extended Kalman filter is employed for dynamic frequency tracking, under stochastic drift and additive noise. The thesis also considers robustness aspects by analyzing the impact of node dropouts - distinguishing between master and slave node failures - and evaluating synchronization stability under varying mobility scenarios. Proper selection of slot duration and signal bandwidth relative to channel coherence time and bandwidth is essential for accurate channel estimation. The results demonstrate that distributed beamforming using software-defined radios is both feasible and effective, provided that appropriate synchronization methods and system constraints are employed.