TCP/IP analysis and optimization over a precommercial live UMTS network

Web browsing optimization over 2.5G and 3G: end‐to‐end mechanisms vs. usage of performance enhancing proxies

“…This leads to inefficient web browsing downloads with long periods of inactivity. In addition, there are other phenomena in WWANs, such as cell reselection in GPRS [9], handover in UMTS [10], link outages and delay jitter [8,23], which may yield undesired interactions with the TCP Retransmission TimeOut (RTO) such as serial timeouts and spurious retransmissions.…”

“…A UMTS network must be able to provide adequate bit rates within its coverage area, offering a peak bit rate of 64 kbps for basic coverage, 144 kbps for extended coverage, 384 kbps in hot spot areas and a theoretical potential peak bit rate of 2 Mbps. Up to 57 kbps uplink and up to 371 kbps downlink IP level available bit rates have been reported from live network measurements in urban environments [10]. Maximum throughput is obtained when large packets are used [10,11].…”

“…Up to 57 kbps uplink and up to 371 kbps downlink IP level available bit rates have been reported from live network measurements in urban environments [10]. Maximum throughput is obtained when large packets are used [10,11].…”

“…Under these conditions, UMTS RTTs are close to fixed Internet RTT values. Measured RTTs fall in the range between 200 and 450 ms depending on the packet sizes [10]. A significant conclusion is that interactive applications will perform better over UMTS than over 2.5G technologies such as GPRS.…”

“…However, recent studies have shown that 2.5G and 3G link layer mechanisms are able to ensure low loss probability to upper layers, especially when the network is properly tuned by the operator [6][7][8]. Actually, the main problem in WWANs is the performance degradation, which is a result of long RTTs in these networks [6][7][8][9][10][11]. Long RTTs have detrimental effects on request/reply protocol patterns.…”