Lecture 5: The sliding window protocol

network utilization (the ratio of throughput to the capacity of the underlying network) is PKT SIZE RTT / C, where C is the network’s capacity, PKT SIZE is the size of a single packet, and RTT is the round-trip time for the path from the sender to the receiver. When we use the variable RTT , we mean the minimum round-trip time excluding queueing delays inflicted by transmissions between the sender and receiver and excluding any packet transmission delays. Let’s plug in some rough numbers for these three parameters to understand how bad the network utilization can be. PKT SIZE is 1448 bytes on the Ethernet link from my office desktop. Let’s pick an RTT of 80 ms to reflect a coast-to-coast US path. The value of C for the path between NYC and a server in SFO is about 370 Mbit/s on my wired network when measured by speedtest.net, The network utilization with these numbers is about 0.04%, which is terrible. In general, the network utilization for the Stop-And-Wait protocol is the least on networks with large RTTs and large link capacity, i.e., long and fat networks (LFN, often pronounced elephen). Clearly, we would like to do better. In today’s lecture, we’ll discuss a protocol called the sliding window protocol with significantly better throughput.