Quantitative analysis of the effects queuing has on a CCID3 controlled DCCP flow

While the Data Congestion Control Protocol (DCCP) shows much promise at becoming a protocol of choice for real-time applications in the future, there are relatively speaking, only a small number of academic papers purporting to its performance and its various nuances. This paper will describe the effects queuing and in particular queue sizes have on DCCP when CCID3 is selected as the congestion control mechanism. From results obtained through the experimentation described in this paper, a clear trade-off between packet loss rates and packet latency values was found to occur when different queue sizes were employed on the experimental network. It was found that employing small fixed sized queues on the network led to lower packet latencies but higher volumes of packet loss as a result of the queue size reaching its maximum threshold more frequently. Alternatively when large queue sizes were used, the number of packet loss events reduced significantly however, packet latency values increased. In addition to showing this impending trade-off empirically, this paper describes ways in which this phenomenon could potentially be exploited to allow DCCP to offer applications with a more tailored form of transportation protocol based on their particular needs.