A drl-based bit rate adaptive UAV video streaming method

Abstract

The invention discloses a DRL-based bit rate adaptive video stream transmission method for an unmanned aerial vehicle. According to the method, the status information of UAV collected by sensors is used as the reference index of ABR, UAV status information includes flight speed, Acceleration, receiver-transmitter distance, and hierarchical processing according to the influence degree of the aboveparameters on the air-ground wireless channel, so that the influence of the UAV state is considered in each video bit rate decision, which is suitable for the situation that the channel state fluctuates violently with the UAV state, and the user experience quality is improved; The LSTM network is applied to the DRL method. The average throughput of the past eight video blocks is processed by the LSTM network, and the characteristics of the throughput time-domain sequence are extracted, so as to better grasp the changes of the channel state, improve the prediction ability of the future channelstate, and make the bit rate selection more correct.

Description

technical field [0001] The invention belongs to the field of unmanned aerial vehicle video stream transmission, and more specifically relates to a bit rate adaptive unmanned aerial vehicle video stream transmission method based on DRL. Background technique [0002] In recent years, the number of video streaming applications based on drones has grown rapidly, such as live video broadcasting using aerial drones. Most video players use an Adaptive Bit Rate (ABR, Average Bitrate) algorithm to optimize video quality. The download and playback of video is carried out in blocks, and each block has several different encoding methods, corresponding to different video quality (bit rate): high-definition, ultra-clear, smooth and so on. What the ABR algorithm needs to do is to automatically select the best bit rate for the next video block to be downloaded according to the observation index on which it is based. [0003] Existing ABR algorithms include buffer-based algorithms, rate-ba...

AI Technical Summary

Problems solved by technology

[0005] Aiming at the defects of the prior art, the purpose of the present invention is to solve the problem that when the DRL-based AB algorithm is applied to the UAV video stream in the prior art, the channel state information CSI is difficult to obtain due to the drastic fluctuation of the channel state before and after the UAV. Estimates are accurate, demodulation is prone to errors, and ultimately lead to technical problems in the decline of throughput

Images