如何在接收器处准确测量视频流吞吐量？

Question

我已经用python开发了一个视频流应用程序。 我一直试图通过以下方式测量接收器（客户端）的throughput (kb/s)：

throughput = [frames's received bytes/s] / [frames receiving time/s]

然而，这种方法有时会给2X或3X分配的带宽（使用tc&；netem工具），特别是对于移动网络中的可变带宽！。特别是，当带宽从一个值更改为另一个值时，测得的吞吐量将增加一倍或三倍

我尝试过的代码：

    frtpPacket,datasize = self.rcvrSock.receiveFRTPpacket()
    frame_start_ts = time.time()
    while prev_frame_no == frtpPacket.frame_no:
        byteReceived = byteReceived + datasize
        # measure packet receiving time
        packet_delay_start =  time.time()
        frtpPacket,datasize = self.rcvrSock.receiveFRTPpacket()
        packetdelay = time.time() - packet_delay_start #in seconds

    accFrame_delay = accFrame_delay + time.time() - frame_start_ts - packetdelay
    if time.time() - start > 1:
        start = time.time()
        byteReceived_in_bits = (byteReceived/accFrame_delay) * 8 # convert to bps
        byteReceived = 0
        accFrame_delay = 0

其中rcvrSock是在类构造函数中初始化的套接字，用于接收帧的数据包。基本上，累积帧延迟和接收字节，然后每1秒计算吞吐量(frames bytes/frames time)

您有更好的方法来应对可变带宽吗

有任何参考或建议吗