"... In PAGE 11: ... We fitted the cumulative distribution function for the file size with a lognormal distribution which we applied in the simulation. Table1 shows the measured parameters. In order to reflect the highly fluctuating connection status of a mobile peer, we describe a mobile peer by an ON/OFF-process.... ..."

"... In PAGE 5: ... We fitted the distribution function for the file size with a lognormal distribution which we applied in the simulation. Table1 shows the measured file sizes and the assumed file appearance probabilities resulting from the conditional probabilities that a peer with a certain storage capacity shares a mobile specific content type. In addition to the mobile peers, we also consider internet peers.... In PAGE 11: ... Figure 6(a) depicts the CCDF of the download time for the four mobile GPRS classes for a popular, cached file of size 858 kB (cf. image category in Table1 ). We see that a higher mobile class, i.... ..."

"... In PAGE 5: ...The following three lists, which are summarized in Table1 . are an attempt to define the nature of P2P, what is new in P2P, and what is not new in P2P.... In PAGE 41: ... In this section, we eval- uate the implications that P2P has for users, developers, and IT department. In Table1 0, we compare P2P with its alternatives. P2P has the following implications for the users of P2P systems and applications: pervasiveness, complexity of use, state of the art, and trust and reputa- tion.... In PAGE 50: ... Centralized systems are costly to own and main- tain and hard to deploy on a wide scale, such as in perva- sive computing. In Table1 1, we summarize various types of P2P systems that will be compared in the rest of the section. We have classified P2P systems into those supporting distributed computing, data sharing, and collaboration, and into plat- forms.... In PAGE 51: ... Table1 2, we compare P2P solutions for distributed com- puting with clusters of PCs, supercomputers and high- end servers, and grids. The earlier systems, such as high- end servers and supercomputers, run either standalone or in intranets.... In PAGE 51: ... Simultaneously, the graphs of security threats are growing and fault tolerance are decreasing, but these plots are omitted from the figure because of the lack of quantitative comparison numbers. In Table1 3, we compare collaborative P2P solutions with two other collaborative solutions: desktop-based computing in an intranet environment and Web-based collaborative solutions. Historically, there seems to be a progressive evolution of connectivity from intranets through the Internet to ad-hoc wireless connectivity.... In PAGE 51: ... However, the security threats are increasing proportion- ally with the increase in connectivity. In Table1 4, we compare the solutions for context shar- ing with their historical alternatives: distributed file sys- tems, and Web-based publishing. Similar to other comparisons, the cost of ownership is reduced, but secu- rity attacks are increased for P2P.... ..."

"... In PAGE 22: ...alues: 2.0, 0.5, 0.125 and 0.0625. The results reported in Table5 refer to a file transfer whose size s is equal to 4 MB and the number n of available resources is equal to 5. Since the load of the P2P network corresponds to the number of concurrent peers that share the available s peer bandwidth during the file transfer of the tagged c peer, when the P2P network load increases the time needed to the file transfer increases.... ..."

"... In PAGE 6: ...2P systems with respect to the five characteristics presented in Section 3.1. These sys- tems are FastTrack, which is the basis of Kazaa, Overnet (including eDonkey), Direct- Connect (DC), and Gnutella. Table1 shows the strong and weak points of these P2P systems, which are based on measurement surveys [3, 4, 6], file-sharing portals (e.g.... ..."

"... In PAGE 7: ...essful P2P systems with respect to the five characteristics presented in Section 3.1. These systems are FastTrack, which is the basis of Kazaa, Overnet (including eDon- key), DirectConnect (DC), and Gnutella. Table1 shows the strong and weak points of these P2P systems, which are based on measurement surveys [3,4,6], file-sharing portals (e.... In PAGE 8: ... Overnet takes a completely opposite approach by offering powerful searching capabilities and queue-based scheduling of downloads with waiting times sometimes exceeding a few days. The big difference between BitTorrent and other systems presented in Table1 is the file-sharing policy. BitTorrent provides a file-level sharing scope by allowing users to download files only if these are already being downloaded.... ..."