Catálogo de publicaciones - libros

To provide scheduling in wireless ad hoc networks, that is both highly efficient and fair in resource allocation, is not a trivial task because of the unique problems in wireless networks such as location dependent and bursty errors in wireless link. A packet flow in such a network may be unsuccessful if it experiences errors. This may lead to situations in which a flow receives significantly less service than it is supposed to, while other receives more, making it difficult to provide fairness. In this paper we propose a QoS-aware fair scheduling mechanism in ad hoc networks considering guaranteed and best-effort flows in the presence of link errors. The proposed mechanism provides short-term fairness for error free sessions and long-term fairness for the erroneous sessions and allows a lagging flow to receive extra service and a leading flow to give up its extra service in a graceful way. It also maximizes the resource utilization by allowing spatial reuse of resource. We also propose a CSMA/CA based implementation of our proposed method.

In this paper, we propose two mechanisms, differentiation by backoff exponent and differentiation by the size of contention window, for IEEE 802.15.4 sensor networks to provide multi-level differentiated services in beacon-enabled mode with slotted CSMA/CA algorithm under non-saturation condition. Mathematical model based on discrete-time Markov chain is presented and analyzed to measure the performance of the proposed mechanisms. Numerical results show that the delicate tuning of throughput could be performed by differentiation by backoff exponent while differentiation by the size of contention window gives more effect for differentiation of throughput.

Wireless sensor networks (WSNs) consist of a large collection of small nodes providing collaborative and distributed sensing ability in unpredictable environments. Given their unattended nature, it is important for the routing algorithm in WSN to deal well with the node failures, resource depletion, and other abnormality. In this paper, an information-driven task routing (IDTR) is introduced for WSN with their specific application to network management. In WSN, a query for the location of the target may be launched from a proxy, which requires the network to collect and transfer the information to an exit node, while keeping the whole communication cost small. The proposed IDTR-based heuristic search algorithm evaluates the expected mutual information of the neighbor nodes, and selects the one that has the maximum information contribution and satisfies the constraint of communication cost as the next hop. In order to avoid the effects of any possible sensor holes, it will calculate the information contribution of all the M hop neighbor nodes to bypass the holes. Simulation results show that IDTR-based heuristic search algorithm achieves the tradeoff between the data communication cost and the information aggregation gain.

In ubiquitous networks, a lot of sensors and RFID readers will be connected to the networks as well as PCs and mobile phones are connected, and huge numbers of transactions are expected to be issued by these devices. Since the loads caused by the transactions are getting increased gradually, it is difficult for system manager to estimate the required performance. The scalable load balancing method is, therefore, expected. This report proposes the scalable load-balancing method in which servers communicate each other, and they manage themselves for load-balancing. Because this method does not need a center server, it can avoid the center-server bottleneck. This system, however, is prone to diverge; the system continues load balancing repeatedly in certain configuration and never stops load-balancing operations. In this paper, we clarify a sufficient condition of the divergence, and prevent the system from the divergence.

This paper proposes a policy-based management framework for self-managed wireless sensor networks (WSNs) called SNOWMAN (SeNsOr netWork MANagement). In WSNs, a number of sensor nodes are deployed over a large area and long distances and multi-hop communication is required between nodes. So managing numerous wireless sensor nodes directly is very complex and is not efficient. The management of WSNs must be autonomic with a minimum of human interference, and robust to changes in network states. To do this, our SNOWMAN architecture is based on hierarchical management architecture and on the policy-based network management (PBNM) paradigm. SNOWMAN can reduce the costs of managing sensor nodes and of the communication among them using hierarchical clustering architecture. SNOWMAN can also provide administrators with a solution to simplify and automate the management of WSNs using PBNM paradigm.

In a large-scale backbone networks, the traffic monitoring system needs to receive a large volume of flow records, so if a single central collecting process is used, it might not be able to process all flow records. In this paper, we propose a method that achieves better scalability by using flow concentrators, which aggregate and distribute flow records. A flow concentrator is located between one or more routers and traffic collectors. The connection methods enable the number of flow concentrators to be adjusted to suit the size of a given network. We propose a reference model for a flow concentrator that is valuable for large-scale networks. We also evaluate the effectiveness of a flow concentrator developed based on these models and the effectiveness of the methods.

Traffic measurement research has focused on various aspects ranging from simple packet-based monitoring to sophisticated flow-based measurement and analysis. Especially, most recent research has tried to address limitations of simple flow-based monitoring by utilizing payload inspection for applications signatures or by identifying target application group based on common traffic characteristics. However, due to highly dynamic nature of the development and the use of the Internet applications, individual simple remedy such as application signature inspection, dynamic port identification, or traffic characterization can’t be sufficient to achieve high precision application-aware monitoring. This is true especially in multi-gigabit high-speed network environment. As the Internet has been evolving from free networks to high-quality business oriented ones, more sophisticated high precision application-aware traffic measurement is required which takes all the factors mentioned above into account. In this paper, we propose our novel traffic measurement methodology to meet such requirements. We considered scalability, cost-efficiency, and performance.

Recent years have seen an increasing interest in end-to-end available bandwidth estimation. A number of estimation techniques and tools have been developed during the last few years. All of them can be classified into two models: rate-based and gap-based, according to the underlying approaches. The difference in characteristics of both models is studied in this paper. and are selected to represent each model and they are evaluated on low speed paths which are typical in today’s local access of the Internet. Finally a hybrid method adopting both rate-based and gap-based approaches is proposed. The hybrid method is compared with and . Simulation proves that the hybrid method yields more accurate results and reduces overhead traffic.

Traffic monitoring is essential for accounting user traffic and detecting anomaly traffic such as Internet worms or P2P file sharing applications. Since typical Internet traffic monitoring tools use only TCP/UDP/IP header information, they cannot effectively classify diverse application traffic, because TCP or UDP port numbers could be used by different applications. Moreover, under the recent deployment of firewalls that permits only a few allowed port numbers, P2P or other non-well-known applications could use the well-known port numbers. Hence, a port-based traffic measurement scheme may not provide the correct traffic monitoring results. On the other hand, traffic monitoring has to report not only the general statistics of traffic usage but also anomaly traffic such as exploiting traffic, Internet worms, and P2P traffic. Particularly, the anomaly traffic can be more precisely identified when packet payloads are inspected to find signatures. Regardless of correct packet-level measurement, flow-level measurement is generally preferred because of easy deployment and low-cost operation. In this paper, therefore, we propose a signature-aware flow-level traffic monitoring method based on the IETF IPFIX standard for the next-generation routers, where the flow format of monitoring traffic can be dynamically defined so that signature information could be included. Our experimental results show that the signature-aware traffic monitoring scheme based on IPFIX performs better than the traditional port-based traffic monitoring method. That is, hidden anomaly traffic with the same port number has been revealed.

It is widely recognized that today’s Internet traffic is mostly carried by a relatively small number of elephant flows while mice flows constitute up to 80% of all active flows at any given moment in time. Although there are many research works that perform structural analysis of flows based on their size, rate, and lifespan, such analysis says very little about temporal properties of interactions among multiple flows originating from different applications. This paper focuses on temporal analysis of flows in attempt to grasp properties and patterns of flows that are related to application and user behaviour and can be captured only in the temporal view of traffic.