Equation-based TFRC

The motivation behind our selection of the Equation-based TFRC (EB-TFRC) [41] among the other AIMD rate control protocols is that EB-TFRC provides smooth rate variation and not too many oscillations. This is because EB-TFRC does not reduce the sending rate by half in reaction to a single packet drop like other protocols and still provides fairness to other competing TCP connections. The primary goal of EB-TFRC is not to aggressively find and use the available bandwidth, but to maintain a relatively steady sending rate while still being responsive to congestion. The main features of EB-TFRC protocol as provided in [41] are: $\bullet$

• Do not aggressively seek out available bandwidth. That is, increase the sending rate slowly in response to a decrease in the loss event rate.
• Do not reduce the sending rate by half in response to a single loss event. However, do reduce the sending rate by half in response to several successive loss events.
• The receiver should report feedback to the sender at least once per round-trip time if it has received any packets in that interval.

Therefore, EB-TFRC best suits multimedia applications because of the overall end-user quality perception. It has been found that users prefer to see video with constant frame rate rather than changing it rapidly depending on the available bit rate [108,124,126]. Like many other AIMD TFRC protocols, EB-TFRC uses a control equation that explicitly gives the maximum acceptable sending rate as a function of the recent loss event rates (that means there is an upper bound the sender should not exceed). The sender adapts its sending rate, guided by this control equation, in response to feedback from the receiver. For traffic that competes in the best-effort Internet with TCP, the appropriate control equation for equation-based congestion control is the TCP response function characterizing the steady-state sending rate of TCP as a function of the round-trip time and steady-state loss event rate, which is given by [41]:

$$T=\frac{s}{R\sqrt{\frac{2p}{3}} +t_{RTO}(3\sqrt{\frac{2p}{8}})p(1+32p^2)}.$$ (81)

Here $T$ is the upper bound on the sending rate in bytes/sec, $s$ is the packet size, $R$ is the round-trip time, $p$ is the steady-state loss event rate, and $t_{RTO}$ is the TCP retransmit timeout value in the congestion state. For more details about EB-TFRC, see [41] and the software implementation available in [5].