Active networks are a novel approach to network architecture that allows user-driven computation to be dynamically introduced within networks. Active networks not only allow a massive increase in service customization capabilities, they also introduce an added complexity for service deployment and management. In this article, the authors propose the use of mobile agents, and present their benefits, as an enabling technology for active networks. A framework for building an agent-based architecture of active networks is also proposed. The authors present different aspects of the framework such as customization encapsulation within agents, customization transfer through mobile agents, and safe customization execution.

Providing predictable and stable Quality of Service (QoS) to the network end users is one of the goals of the next generation Internet. To solve different problems related to QoS, the Internet pricing has been researched. This paper proposed a multi-service control framework based on pricing and charging technologies. It consists of three fundamental blocks: intelligent agent (IA), pricing broker (PB) and local pricing agent (LPA). The intelligent agent provides the TCP-like pricing based traffic control at the end users. The local pricing agent is used to implement hybrid-pricing algorithm to make the service price as an indicator of the network status. At the network edge node, it also contains traffic classification mechanisms to provide service differentiation. But the pricing broker controls the policies. It is also responsible to maintain and exchange the price information for the end users and neighbor domains. A simulation has been done in a simple prototype with the hybrid-pricing algorithm and the price based classification. Simulation results show that it can provide service differentiation and maintain the service quality as well. Therefore, the proposed framework provides a simple, flexible way to support multi-service control and improve QoS over the networks via pricing technology.

Needs of Information Security in Multicast is increased. As clients join or leave a specific service group, Backward and Forward Secrecy problem occurred. Solving this problem, service group will make a re-key operation periodically. But because of this operation need translation frequently so it may have a bad influence to Real time property, which needs minimum bandwidth requirement. In this paper, we proposed a Group Key Management System, which is comprised of two levels, KD (Key Distributor) subsystem and subgroup, for managing encryption key. A KD (Key Distributor) subsystem is composed of SKDs (Subgroup Key Distributor) and TKD (Top-level Key Distributor). A SKD manages a encryption key of a subgroup. A TKD generates a KD group key that is a encryption key used in a KD group and transmits it to SKDs with safety. Subgroup consists of hosts in Multicast group. Hosts and a SKD share a encryption key, a subgroup key. This key is generated by a SKD and cannot be disclosed outside of the subgroup. As a result, a load of key management can be distributed into many KD so that the overhead of key translation can be decreased, which is needed at each stage of Multicast traffic. In joining and leaving a Multicast group frequently, a group key is distributed only in a specific subgroup. Therefore the overhead needed to redistribute a key can be decreased. By reducing overhead from security service, we expect to satisfy real time property.

Mobility support on the network layer is of special importance, as the network layer holds together the huge Internet with common protocol IP. Although based on possibility different wireless or wired technologies, all nodes of the network should be able to communicate. Therefore, mobile IP (an extension of the classical IP) has been designed which enables mobility in the Internet without changing existing wired systems. However, mobile IP leaves some points unsolved. Especially, if it comes to security, efficient of the packet flow that is called triangular routing. Especially triangular routing can cause unnecessary overhead for the network. Furthermore latency can increase dramatically. This is particularly unfortunate if two communicating hosts are separated by transatlantic links. In order to this problem, many methods like IPv6 and ROMIP are proposed. But these methods have limitation. In other words, they have not compatibility because of needing modification or original IP scheme. Especially ROMIP is very complex and the overhead of control message and processing are serious. The problem of inconsistency of Binding caches may occur tool. We therefore propose and analyze the DRMIP (Direct Routing Mobile IP) which do not modify IP source needed in the sender, is compatible with IP and Mobile IP.

Third Generation Wireless Systems are rapidly moving towards the development after the successful completion of prototypes, demonstrating the core functionality. CDMA-2000 is a third generation mobile wireless system on which integrated voice and data applications will be supported. In an IS-634 wireless access network, the Selection Distribution Unit (SDU) performs the frame selection and distribution of traffic frames carrying voice, data, and signaling between a mobile station and the wireless network. In this paper we present architecture of a SDU based on a network processor and analyze the performance of CDMA-2000 voice traffic on it. We also include the performance results obtained on a general-purpose processor based SDU. The performance results indicate that a network processor based SDU has significant better throughput performance over the SDU unit equipped with general-purpose processors.

The discussion will cover the new initiative to create a new MAC layer standard for resilient packet rings: IEEE 802.17 RPR. The key aspects of the presentation will include a preliminary address of the Metro Area Network today and the current networking technologies such as SONET/SDH which are not optimized to carry IP traffic over Metro MAN. The next segment will cover the options which could change the traditional and expensive layered networking model, and address the real benefits of marrying several technologies like Ethernet, SONET/SDH and IP into one technology. The next part of the discussion will detail the technical advantages a new MAC will bring to the services providers. Lastly a summary of the view and strategy about the acceptance and deployment of this new technology in the next 12 months, specifically, now one defines and develops standards for a Resilient Packet Ring Access Protocol for use in Local, Metropolitan, and Wide Area Networks for transfer of data packets at rates scalable to multiple gigabits per second; specifically address the data transmission requirements of carriers that have present and planned fiber optic physical infrastructure in a ring topology; and, defining and developing detailed specifications for using existing and/or new physical layers at appropriate data rates that will support transmission of this access protocol.

Transparent optical networks are replacing SONET/SDH in metropolitan and regional areas. They must offer protection, switching/routing, and performance monitoring capabilities. These functionalities can be provided by optical switching matrices that switch either fibers, groups of wavelengths, or single wavelengths. These switches can be used for rapid fiber or service provisioning, and for enhanced networking functionalities. Several applications are discussed hereinafter. Examples are fiber cross connects, flexible optical add/drop multiplexers, shared optical protection rings, and shared performance monitoring.

The design of a survivable optical network has been extensively studied, but most of the work has confined to wavelength level path or link protection and restoration, while the other work has relied on fiber protection/restoration. Each scheme has its own pros and cons. Based on these existing schemes, we have presented a hybrid protection scheme (HPS), which divides the optical layer into wavelength layer and fiber layer, to provide more effective survivability for multiple fiber mesh-based WDM networks. In this paper, we will further extend HPS concepts to allow local wavelength level recovery in order to resolve breakdown in few wavelengths by sharing the resources from fiber level recovery. A detailed survivable node architecture design is proposed, and how this node architecture provides network survivability against various failure scenarios is presented. The performance of multifiber network and single fiber network is also compared; finally, the various protection schemes are evaluated.

Recently there are considerable amount of research about the automatic control and provisioning in all optical networks. One of the critical issues is how to provide effective lightpath provisioning to improve network performance, such as blocking probability and decision time. Depending on the network topology, configuration, and administration policy, a distributed or centralized control scheme can be employed to manage the routing and signaling. In a distributed control scheme, each node exchanges information with other nodes, but performs routing and signaling independently from other nodes. On the other hand, in a centralized scheme, each node communicates with a central controller and the controller performs routing and signaling on behalf of all other nodes. Intuitively, the centralized scheme can obtain a lower blocking probability since the controller has the complete resource availability information. We have studied the two schemes through emulations, determined the signaling and processing overheads and quantified the conditions that favor one approach over the other.

Many designs and topologies have been introduced for optical multistage interconnections networks (optical MINs). One of the promising topologies that is regular and can be used in large optical switches is the family of networks that have the Quartet property. The Quarter property is a new property of a class of equivalent MINs. This class of networks includes Omega Network, Flip network, Baseline network, Reverse Baseline network, Indirect Binary n-Cube network and some other networks which are topologically isomorphic to them. Some MINs with the Quartet property such as Omega Network and Baseline Network have topologies that recursively can be expanded or reduced with an exponential growth of a factor of four. We are going to use this characteristic of these networks in the construction of our optical switch. The Quartet topology is a unique MIN in which a fiber channel is used instead of a set of wires. Through advancements in fiber optic channels with high bandwidth, this topology is a perfect candidate for optical multistage interconnections networks and optical switches. This paper presents the design for a Quartet switch with recursive growth based on an Omega network.

Traffic engineering must be concerned with a broad definition of service that includes network availability, reliability and stability, as well as traditional traffic data on loss, throughput, delay and jitter. MPLS and Virtual Private Networks (VPNs) significantly contribute to security and Quality of Service (QoS) within communication networks, but there remains a need for metric measurement and evaluation. The purpose of this paper is to propose a methodology which gives a measure for LSP ( Lfew abel Switching Paths) metrics in VPN MPLS networks. We propose here a statistical method for the evaluation of those metrics. Statistical methodology is very important in this type of study since there is a large amount of data to consider. We use the notions of sample surveys, self-similar processes, linear regression, additive models and bootstrapping. The results obtained allows us to estimate the different metrics for such SLAs.

The increasing demand of new services and applications is pushing for drastic changes on the design of access networks targeted mainly for residential and SOHO users. Future access networks will provide full service integration (including multimedia), resource sharing at the packet level and QoS support. It is expected that using IP as the base technology, the ideal plug-and-play scenario, where the management actions of the access network operator are kept to a minimum, will be achieved easily. This paper proposes an architecture for access networks based on layer 2 or layer 3 multiplexers that allows a number of simplifications in the network elements and protocols (e.g. in the routing and addressing functions). We discuss two possible steps in the evolution of access networks towards a more efficient support of IP based services. The first one still provides no QoS support and was designed with the goal of reusing as much as possible current technologies; it is based on tunneling to transport PPP sessions. The second one introduces QoS support through the use of emerging technologies and protocols. We illustrate the different phases of a multimedia Internet access session, when using SIP for session initiation, COPS for the management of QoS policies including the AAA functions and RSVP for resource reservation.

This paper describes a framework for the design of next generations of multicast services. The framework is based on active and programmable networks, distributed objects and distributed control algorithms with quality-of-service (QoS) constraints. Many of the Internet applications require variable (dynamic) QoS guarantees from networks. This is the case of distributed virtual environments that support the execution of simulation applications, or audio-video communication applications over the Internet. With the development of active and programmable networks, distributed object platforms and system technologies, it becomes feasible to provide multicast services that can be controlled to adapt the allocation of network resources to QoS variations. Multicast services based on the framework described in this paper can be used by application software components to manage network resources, to bring modifications to computations associated to network nodes, and to easily implement distributed control algorithms and services that can take into account variable QoS requirements expressed by applications. The paper focuses on the principles that govern the design and implementation of multicast protocols based on this framework, and the study of some of their major benefits. This work is partially supported by the European Commission - IST research program, under agreement IST-1999-10561: project FAIN.

The Data Over Cable Service Interface Specifications (DOCSIS) of the Multimedia Cable Network System (MCNS) organization intends to support IP traffics over HFC (hybrid fiber/coax) networks with significantly higher data rates than analog modems and Integrated Service Digital Network (ISDN) links. The availability of high speed-access enables the delivery of high quality audio, video and interactive services. To support quality-of-service (QoS) for such multimedia applications, it is important for HFC networks to provide effective medium access and traffic scheduling mechanisms. In this work, we consider an HFC network that has a shared upstream channel for transmissions from stations assigned with different service priorities to the headend. We first present a multilevel priority collision resolution scheme with adaptive contention window adjustment. The proposed collision resolution scheme separates and resolves collisions for different classes of critically delay-sensitive and best effort traffics, thereby, achieving the capability for preemptive priorities. To enhance the performance of the proposed scheme, we adopt a novel methodology in which the headend dynamically selects the optimal backoff window size according to the estimate of the number of contending stations for each priority class. A traffic scheduling policy with multiple priority queues is also employed in the headend to schedule data transmissions. This scheduling strategy is used to satisfy bandwidth requirements for higher priority traffics. Simulations are conducted by using OPNET. We present a set of simulation scenarios to demonstrate the performance efficiency of the proposed scheme.

Colorado State University (CSU) operates a radar at Greeley, Colorado. The CSU-VCHILL (Virtual-CHILL) project is aimed at developing and implementing protocols for providing the raw radar data to researchers in real-time over the internet. The Next Generation Internet has features, which could be used to transfer raw time-series data generated by the radar. In this project we have developed a network transfer application with TCP as the underlying protocol. This network transfer application has been successfully tested to provide throughput of 90 Mbps over a 100 Mbps link and 300 Mbps over gigabit link. We have also implemented a software digital signal-processing module. A network transfer application with User Datagram protocol as the underlying protocol was also developed. The performance of the software digital signal processing unit and the TCP version of the application to transfer the data across the network meet the requirements in most areas.

Security and performance are probably the top two concerns of web hosting service providers. As available bandwidth of a hosting service is approaching Giga-bits-per-second, low throughput of a single firewall quickly becomes the bottleneck. Constructing a load-balancing cluster of multiple firewall devices seems to be an effective solution. In this paper, we first present a proof-of-concept firewall cluster using web load balancing switches. Our test cluster works; but has major limitations. First, the cluster set-up is too complex to be manageable in a large-scale deployment. Furthermore, the firewall cluster works only in a local area network. It does not work across the wide area network where asymmetric routing is possible. Based on these findings, we propose two novel approaches. The first approach introduces a Firewall Cluster Control Protocol (FCCP) for routers to direct network flows to the appropriate firewall device for processing. FCCP simplifies the implementation of firewall clusters by eliminating the load balancing switch requirement. The second approach, called Stateful Packet Forwarding (SPF), allows firewall devices in a cluster to discover the 'owner' of a network flow when asymmetric routing occurs. SPF can be potentially used in a geographically distributed firewall cluster.

Radar design and management concepts are presented for a networked environment to reduce the cost and dramatically enhance the performance of radar systems. The benefits attained by moving from a large centralized radar to a distributed cluster of smaller radars extend far beyond the reduction in cost and the increase in reliability. Additional advantages of multiple radar operation exploiting the ubiquitous networking technology are also presented. In addition to providing the impetus for novel radar algorithms and applications, the proposed system will also a yield a testbed for challenging networking research to develop fast protocols dealing with very high bit rate applications relying on multiple streams of time-critical data.

Limited attention has been paid to the interactions of service restoration protocols that operate during a fiber cut to restore connectivity between communications equipment. Historically, restoration protocols were deployed only at the SONET layer in telephony networks. SONET frames that carried voice-grade signals such as a T1 or T3 would be redirected to protection path over architectures that supported bidirectional communication. With the advent of new communication technologies such as ATM, IP and WDM, a cable cut can affect multiple routing processes at each of these layers even if a particular network region only supports a few of these technologies. For example, restoration processes at an arbitrary layer in adjacent networks might trigger if lower-layer protocols don't finish within specific deadlines. With the growth of data traffic and a wide range of service offerings, the telecommunication networks of the future are growing more complex, requiring multiple interactions between software systems. Failures will be difficult to pinpoint and the cooperation of the repair processes will be key to ensure that services traversing multiple networks are not interrupted.

In this study, business benefits for this Virtual Private Network (VPN) and protocols, techniques, equipments used in this VPN are reported. In addition, our design experience for fault tolerance, security and network management and administration on this VPN are showed. Finally, the issues for future planning of this VPN is addressed.

Recently, wavelength division multiplexing (WDM) technology has been widely employed to increase the capacity of existing SDH/SONET self-healing rings. As wavelengths are no longer such precious resources, SDH/SONET add-drop multiplexers (ADMs) become the dominant cost factor in network deployment. Traffic grooming studies how to intelligently arranging the placement of ADMs on wavelengths to reduce the number of ADMs required to support certain traffic patterns. In this paper, we address dynamic traffic grooming in interconnected WDM unidirectional rings. An optically interconnected dual-homing strategy is adopted to implement optical layer survivability. A genetic algorithm (GA) based approach is proposed for static traffic grooming. Another GA based approach is proposed for combining different topologies to support a given traffic set. Numerical results were reported on ADM savings.

An all-optical Manhattan Street Network (MSN), in which the data portion of a packet is maintained in an optically encoded format from the source to the destination, can be achieved by augmenting MSN with optical buffering. Augmenting MSN with optical buffering improves its performance significantly and reduce its complexity by reducing or even eliminating the need for optical-electrical conversions. The performance of an MSN with optical buffering is compared with that of an MSN in which only store-and-forward routing mechanism is used. The network is based on a novel 6x6 optical switching node.

An application layer based QoS mechanism is proposed so that QoS to IP Traffic is provided solely based on the individual requirements of the applications. The mechanism of communication between application layer and IP layer to provide information of the QoS requests, the marking model of IP packets, and the evaluation of end to end performance in terms of standard QoS parameters are addressed. We have proposed 5 categories of Internet traffic based on their QoS requirements. We also have carefully studied the accommodation of popular TCP based applications like FTP and have minimized its interaction with other UDP applications. Finally, mapping is presented for proposed Internet traffic categories in a way that each of them achieves its QoS requirements.

The deployment of wavelength division multiplexing has increased the demand for optical switching solutions in the core network. Recent developments in a number of emerging optical switching technologies are now making it possible to build new networks where switching of multiwavelength data can be carried out in the optical domain. This development is leading the way to a future core that is mostly optical. This paper looks at the evolution to optical switching and gives a brief overview of the situation of a number of optical switching technologies.