### Table 2: Performance of DHT on Static Networks. Results are the means of three simulations. ability of the DHT to hold data written to it. To measure the storage load on nodes, we examine the maxi- mum number of events stored at any node, to capture the worst- case required storage; and the mean number of events stored across all nodes in the network, to capture typical storage re- quirements. We measure the communication load on nodes by counting the mean number of messages forwarded by a node in a refresh interval, and the mean number of refresh messages for- warded by a node in a refresh interval; these message counts are averaged across all nodes and refresh intervals in a simulation.

### Table 2.1: Comparison of the five previously described DHT algorithms according to [12]. N stands for the number of nodes in the network, d for the number of dimensions in CAN

2006

### Table (DHT) instead of SIP servers for user location and user registration [16], [17]. Distributed Hash Tables have been developed to reliably store and retrieve content in a distributed network3.

### Table 3: Performance of DHT. Stationary nodes, varied fraction of nodes alternate between up and down states. Results are the means of eight simulations.

"... In PAGE 10: ... All the results we present in this section are for networks of 100 nodes. Table3 shows the performance of DHT under a failure model where a configured fraction of nodes selected uniformly at ran- dom alternate between failing and restarting. When a node fails, it loses the contents of its database; it only reacquires its database contents upon returning to operation and receiving refreshes from neigbhors.... ..."

### Table 1: Comparison of various protocols. The cur- rent size of the network is n.

2003

"... In PAGE 1: ... Thus DHT routing topologies face two conflicting goals: fast lookups but small state. Table1 sum- marizes the trade-offs offered by various DHT topologies. All the protocols are scalable and handle dynamic networks.... ..."

Cited by 36

### Table 1: Comparison of various protocols. The cur- rent size of the network is n.

"... In PAGE 1: ... Thus DHT routing topologies face two conflicting goals: fast lookups but small state. Table1 sum- marizes the trade-offs offered by various DHT topologies. All the protocols are scalable and handle dynamic networks.... ..."

### Table 1: UsenetDHT reduces the bandwidth and storage requirements of hosting a Usenet feed in proportion to the fraction of articles read (r) and the number of servers in the network (n), respectively. This table compares the cur- rent transfer and storage requirements per day for a full Usenet feed in both systems, where b represents the total number of bytes for articles injected each day. A single peer is assumed

### Table 2. Statistic of DHT objects stored in ConDHT.

### Table 1: Parameter settings for the uni ed discrete transformation architecture, where RefXDFT (k)g and ImfXDFT (k)g denote the the real part and the imaginary part of the DFT, respectively. ck is the scaling factor used in the DCT/DST and their inverse transforms.

"... In PAGE 7: ... As an example, the discrete Hartley transform (DHT) can be represented as XDHT (k) = XC(k) + XS(k) (16) by setting L = N; = 1 pN ; !k = ?k N ; and k = ?!k: (17) The other example is the Modulated Lapped Transform (MLT) [18] which can be computed by XMLT (k) = ?sk[ XC(k + 1) + XS(k) ] (18) for k = 0; 1; : : : ; N ? 1, where the scaling factor sk = (?1)(k+2)=2 if k is even, and sk = (?1)(k?1)=2 if k is odd, and the parameter settings are L = 2N; = 1 p2N ; !k = k 2N ; and k = 2 (k + 1 2): (19) (16) and (18) are referred to as the combination functions which describes how to combine XC(k) and XS(k) together to obtain the desired DT coe cients. The parameter settings and the corresponding combination functions for most DTs are listed in Table1 . The corresponding IIR-based uni ed DT architecture can be... In PAGE 9: ...4 performs the block transforms in an SIPO way. The interconnection network will be con gurated according to the combination functions de ned in Table1 . The detailed settings of the interconnection network used in this paper (Type I-IX) are described in Table 3.... In PAGE 9: ... After the last datum enters the uni ed module array, the evaluations of XC(k) and XS(k) in (14) and (15) are complete. Then the interconnection net- work will combine the module outputs according to the combination function de ned in Table1 , and the transform coe cients can be obtained in parallel at the outputs of the network. At the same time, the host processor will reset all internal registers (delay elements) of the programmable modules to zero so that the next block transform can be conducted immediately.... In PAGE 14: ... There are two parts inside the interconnection network: One is the summation circuit to combine the even and odd outputs of the array. The other is the circuit to perform the combination function de ned in Table1 . In real implementation, we can either add one summation circuit so that the switch settings for the DT in Table 3 can still be used, or we can de ne new switch settings by merging these two circuits together.... ..."

### Table 1: Comparison of Various DHTs Comparison Criteria Overlays

"... In PAGE 5: ... There have been attempts to devise a common Application Programming Interface (API) for different DHT algorithms [32]. Table1 illustrates differences in various DHT facilities. Chord, as explained above, arranges the network in circular space.... ..."