On a multimedia digital home network, an audio player may be separated from the video player and both of them are connected by wireless interface, like 802.11 series. Due to forwarding, rewinding, or packet loss, it may make the video player and the audio player asynchronous. Based on the original UPnP AV architecture, a synchronous mechanism is proposed in this paper. With exchanging both status of the video player and the audio player in UPnP messages, the synchronization is accomplished. A prototype was also implemented to verify its feasibility.

It is used to identify users' connections by using the source IP addresses in the traffic log inside a firewall. In the case with pre-assigned IP addresses, the mapping relationship between IP addresses and users is fixed. However, an IP address can be counterfeited or abused. Therefore, an evil user can do any wicked network behavior with others' IP addresses when they get off work or take off. Users who originally own the source IP addresses become victims. In case with dynamically assigned IP addresses, it is more difficult for network managers to identify the user of a connection. In a wireless environment, although some authentication mechanisms may be applied, they are, however, used just to control the wireless access right, instead of to authorize the network behavior of a user. It is much more difficult for network managers to identify the user of a connection. In this paper, a user-level packet filter for wireless/wired networks is proposed. Any user must be authorized and register with the user's current IP address such that the user can use the Internet service. The proposed user-level packet filter solves IP address counterfeit problem. Little overhead is added but without degrading the performance too much.

Due to security reasons, many companies need firewalls to filter some mistrusted applications, like FTP or P2P software. However, some applications may hide themselves with some well-known application ports like HTTP port 80 such that some firewalls cannot distinguish mistrusted applications from well-known applications. As a result, firewalls require high performance classification engines that can efficiently inspect layer-7 contents to recognize mistrusted applications. This paper analyzes the layer-7 classification module in Linux Netfilter, the L7filter package, and proposes an alternative implementation to improve the performance of L7filter. The throughput of the proposed method can remain high even in heavily-loaded network environments. The performance of the proposed method is justified by the Spirent SmartBits 6000 testing equipment whose traffic generation speed can achieve gigabit wire-speed.

The deployment of the next-generation Internet Protocol, IPv6, has progressed rather slowly. Although some commercial products have been incorporated with parts of functions of IPv6, it is still difficult to have a revolution from IPv4 to IPv6 at one night. Since the current users on line must be cared, the migration from IPv4 to IPv6 must begin from integrating both IPv4 and IPv6 networks. In this paper, we propose a hierarchical routing architecture to integrate both IPv4 and IPv6 networks. It improves encapsulation overhead of tunneling based methods and reduces consumption of IPv4 addresses in header-translation based methods.

We consider the scheduling problem in time-division multiplexed (TDM) switching systems. In previous works, the interdependence between traffic demands in two consecutive frames is neglected, and scheduling algorithms found up to now have time complexities O(N/sup 5/) or O(N/sup 4.5/), where N is the switch size. However, in many applications like voice or video communications, if a source transmits a packet to a destination in a frame, it is highly probable that it will also transmit a packet to the same destination in the next frame. So it is not necessary to schedule incoming packets for every frame if we can preserve all the switching patterns for the nearest scheduled frame and update the patterns appropriately according to the changes of traffic demands. The adaptive algorithm proposed in this paper assigns time slots to packets according to the changes of traffic demands. This algorithm has the worst case time complexity O(N/sup 2/L), where L is the TDM frame length. Comparing the time complexity of the adaptive algorithm with those of previous scheduling algorithms, the adaptive algorithm can perform better than previous scheduling algorithms when N is large and/or L is small. Since traffic demands in consecutive frames are expected to be interdependent in many applications, the proposed algorithm may offer as an efficient alternative for scheduling time slots in these applications.>

The authors propose a cell scheduling algorithm to increase the throughput of an asynchronous transfer mode (ATM) switching system with services of diverse priority classes. The maximum throughput is increased up to 0.949. The efficiency (output trunk utilization)/(input trunk utilization) is almost equal to 100%, and is almost independent of the switch size and the traffic load. The cell scheduling algorithm is shown to be able to provide high throughput not only for higher priority classes but also for lower priority classes. The mean cell waiting time of the highest priority class are transmitted as soon as possible. In the switching system implemented with the present cell scheduling algorithm, it is sufficient for a sorting network to establish input/output paths through it simultaneously without conflicts and it is not necessary to append a routing network to a sorting network.>

The design of a parallel theorem prover for first-order logic is described. The parallel theorem algorithm is based on the divide-and-conquer strategy. The concept of restricted substitution is used to reduce the number of ground clauses generated during the operation of this theorem prover. In this manner, the ground clause set generated by the theorem prover will be much smaller than that generated directly by Herbrand universe.>

Consideration is given to the scheduling problem for avoiding output conflicts in time-division multiplex (TDM) switching systems. In many applications, if a source transmits a packet to a destination in a frame, it is highly probable that it will also transmit another packet to the same destination in the next frame. So it is not necessary to schedule incoming packets for every frame if all the switching patterns for the nearest scheduled frame can be preserved and the patterns updated appropriately according to changes in traffic demand. An adaptive algorithm of a lower time complexity which assigns time slots to packets according to the changes in traffic demands is given. This algorithm has the worst case time complexity O(N/sup 2/L), where L is the TDM frame length and N is the switch size. Simulation results show that the algorithm is efficient for interdependent traffic and for random traffic.>

The authors consider the output contention problem with a view towards increasing the throughput for asynchronous transfer mode (ATM) switching systems. A cell scheduling algorithm for increasing the throughput is proposed. The maximum throughput is increased up to 0.957. The efficiency (output trunk utilization/input trunk utilization) is almost equal to 100% and is independent of the switch size and traffic load. A switching system implemented with this cell scheduling algorithm is also proposed. The switching network usually consists of a sorting network followed by a routing network. Here, it is sufficient for a sorting network to establish input-output paths through it simultaneously without conflicts, and it is not necessary to append a routing network. In addition, a parallel mesh-connected architecture of a component of the switching system is proposed to speed up the cell scheduling of the system. Consequently, this approach can offer an effective alternative to ATM switching systems.>