Table 1. Measured ping times (32 bytes) Technology Bandwidth (down/up) Mean Std In this paper we present results from a measurement GPRS 80/40 kbit/s 488 ms 146 ms campaign evaluating the web performance of di'fferent mo-EDGE 240/120 kbit/s 504 ms 89 ms bile Internet access technologies with and without caching. UMTS 384/128 kbit/s 142 ms 58 ms We measured in GPRS, EDGE, UMTS and HSDPA radio HSDPA 1800/384 kbit/s 91 ms 43 ms access networks ofa mobile operator in Austria. The opera-ADSL 1000/256 kbit/s 10.9 ms 0.8 ms tor has implemented a web-proxyfor performance improvement. To evaluate the impact on the different technologies nation with flat rate contracts has introduced mobile Interwe ran the setup with and without the web-proxy. Wedefined net access to a large population. Complete UMTS coverage the download time ofweb pages as the performance index for our work. We developed a tool to capture this parameter is esve for the operatort g in raefore,i oer and derived the index for a set ofweb sites frequently used to secsime o E oErators m ate teir Gac by the customers in the network. Beginning with an over-GPRS equipment to EDGE (Enhanced Data rates for GSM bylthcurstomwe winlpth senetwork Begain analysisfor over-l Evolution), which provides three times the bandwidth of calls.W comparison w e wimprest etaled analysistem for spfeia GPRS. Note that the RTT in this case still remains at the cases. We evaluated the impact ofthe proxy system to Ffllerlevel of GPRS, see Table 1. HSDPA (High Speed Downlink ent site configurations using the performance index. Finally Packet Access), an evolution of UMTS, offers more than we compared the performance numbers for HTTP with re-1'Mi/ atavr o ea f 9m.TeRTvl sults colletedfromFTPtransfers.1.8Mbit/s at a very low delay of r-. 90mrs. The RTT values we used in this paragraph are well known in literature, for example see [1]. However, we also measured these values using ping requests, contacting the Gi interface of the 1 Introduction GGSN. The results for 100 ping packets with a payload size of 32 bytes is shown in Table 1. The fast growth of the Internet has also influenced the As we have shown in [2] the IP traffic in the network evolution of mobile networks. The introduction of GPRS of a mobile operator, in terms of volume, is dominated by (General Packet Radio Service) to the mobile market was HTTP traffic, measured in bytes over TCP port 80. The a step to cover the growing demand for packet switched intention of this paper is to find out how the different access services. This was the first move from the pure voice oritechnologies with their different parameters perform to the ented GSM (Global System for Mobile Communications) perspective of a web user. network to a system supporting a wide range of new serThe rest of this paper has been organized in the followvices. With a throughput of -80kbit/s and a RTT (Rounding way. In Section 2 we show the measurement setup, inTrip Time) of -500ms GPRS was not competitive to wired cluding a flowchart for the coded software and a desc...

No abstract

No abstract

No abstract