Time-varying connectivity is one of main challenges faced by controlling a team of unmanned aerial vehicles (UAVs) in the multipath fading channel, incurring low accuracy and significant convergence time of formation control law. To address this issue, in this study, a topology optimised based decentralised consensus is developed for controlling a multi-UAV system in a multipath fading channel, in which a formation structure reconfiguration scheme is proposed as well as a transmission power allocating algorithm to guarantee the control accuracy in a limited convergence time. In particular, the objective function for topology optimisation is well-designed by considering the second eigenvalue of Laplacian matrix of topology as a feasible index of connectivity degree. To improve the efficiency of information transmission, a specified consensus protocol is proposed with well-tailored packet format and signalling procedure for control messages. Through the comparative simulation results, the proposed consensus can achieve high convergence accuracy and less convergence time in a multipath fading channel, indicating high resilience of the proposed protocol under a multipath fading channel.