User perceived latency is the most prominent performance issue influencing the World Wide Web (www) presently. Hyper-Text Transfer Protocol(HTTP) and Transmission Control Protocol (TCP) have been the backbone of web transport for decades, thus received a lot of attention recently due to end-to-end performance degradation in mobile environments. Inefficiencies of HTTP and TCP strongly affect web response time mainly in resource limited devices. HTTP compression reduces some of the burden imposed by TCP slow start phase. However, compression is still an underutilized feature of the web today [1]. In order to fulfill the end user expectations, we can optimize HTTP to improve Page Load Time (PLT), low memory usage and better network utilization. SPDY, a web latency reducing protocol and HTTP pipelining are a recent proposal to provide faster information exchange over web. Through the course of this work, we present a comprehensive study of new approaches to reduce mobile web latency. At first, we measure the PLT after implementing SPDY, HTTP and HTTP pipelining. Secondly, we also analyze the performance of these protocols after tuning the network parameters like bandwidth and round-trip time (RTT). Finally, we compare the performance of HTTP and other latency reducing protocols. We have conducted all experiments over DummyNet under user-configured network conditions. We critically discuss the challenges of shifting from HTTP to these latency-reducing protocols.

This diploma thesis focuses on theoretical and practical comparison of webpage transport protocols such as HTTP/HTTPS (Hypertext Transfer Protocol) v1.1, SPDY, HTTP/2 and QUIC using UDP (User Datagram Protocol), TCP (Transmission Control Protocol) and SCTP (Stream Control Transmission Protocol) as transport protocols. This work also contains design and manual for practical laboratory tasks on which can students verify theoretical assumptions. These tasks compares protocols in different conditions such as packet loss, latency and jitter.

Orientador: Nunzio Marco Torrisi / Dissertação (mestrado) - Universidade Federal do ABC. Programa de Pós-Graduação em Ciências da Computação, 2014