Search

Usage of data networks encompasses not only the traditional data applications but also newer applications such as realtime audio/video streaming, voice over TCP/IP, realtime transactions, asynchronous messaging and other batch transactions over digital networks. Each application has different capacity and quality of service (QoS) requirements. Each is affected differently by network reliability and speed. Major network providers have already started efforts to accommodate the QoS demand generated by these applications. Providers charge differently for the capacity they sell. For example, Internet service providers (ISPs) offer a combination of fixed price for a fixed maximum bandwidth (allyoucan send) or pay per hour (or minute) for again a maximum bandwidth. Wireless phone companies charge for text messaging based on bytes sent. Certain calling plans offer different per minute charges for phone conversations depending on when the call is placed. A firm uses data networks to perform and support business operations, which we will call tasks. For example, a video conference is a type of task, so is a remote web site update. The obvious commonality is that both tasks require network resources. A network resource is characterized by its capacity (bandwidth and duration) and QoS. We assume that the contract signed between the firm and the provider specifies the amount of bandwidth and quality guaranteed at a given time. Duration specifies the period during which the resource is available. There are two types of costs associated with using data networks. The first one is the resource (i.e., bandwidth, capacity) acquisition cost. The second is the opportunity cost incurred due to insufficient quality realized in performing certain tasks such as video conferencing. Given a heterogeneous set of tasks and resources that differ in bandwidth and quality requirements, it might be in the firm's best interest to use either multiple providers or sign multiple contracts with different bandwidth and quality requirements. For example, while resources with lower quality can be used for data applications, more expensive resources might be utilized for realtime applications with high QoS requirements. Therefore, the problem that the firm has to solve is a cost minimization problem that reflects a trade off between the cost of acquiring resources and the opportunity cost of degradation in realized quality of tasks performed. The quality of service of the resource affects the customer in two ways. First, size-fixed tasks might be delayed beyond acceptable deadlines. Second, the realized quality of a time-fixed task such as a video conference might be unacceptable creating an opportunity cost for the customer. In general, opportunity cost reflects the importance of a task. The more important the task is, the higher the penalty for not achieving desired targets (such as audio or visual quality). The decision maker will minimize the total cost by considering the tradeoff among these costs when assigning tasks to resources. In our study we formulate a cost minimization problem subject to QoS and capacity constraints. We consider the taxband pricing scheme suggested by Courcoubetis et al. (2000a) in which each supplier has a convex piece wise linear cost function for each resource offered. Courcoubetis et al. (2000a) claim that taxband pricing reduces bursty traffic since customers are likely to reduce such demand to avoid paying more. Suppliers in general would prefer having more customers with less capacity demand rather than fewer customers with high capacity demand. Given this pricing structure the customer has to decide how much capacity to acquire and how to allocate tasks. For tractability we relax the due date constraints and assume that all tasks and resources are available at time zero. We also assume that realtime applications have desired transmission rates and any deviation from that creates an opportunity cost for the customer. In spirit, this is the same objective function used in Kasap et al. (2007) that trades the quality cost with the capacity cost but the cost structure for capacity is quite different in this problem. In this study, firstly, we present taxband pricing. Secondly, a formulation and a solution procedure based on GBD are described. [ABSTRACT FROM AUTHOR]

Underwater Sensor Networks (USNs) can improve naval defense, earthquake/tsunami forewarning, water pollution detection, ocean life monitoring systems, etc. Stationary Underwater Sensor Networks are ideal for securing or monitoring a fixed target region, e.g. monitoring oil drilling platforms for spill detection, harbor entrances for surveillance, ocean bottom for seismic activity observation, etc. On the other hand, mobile untethered Underwater Sensor Networks are flexible and better alternatives for short term exploration of moving targets. For instance, untethered, free-floating underwater sensors can track a chemical spill or a pollutant that may be dangerous to human health or sea life. In a sensor network, sensor nodes collect data from their surrounding and tag these data, in order to transmit them to a more powerful node for processing. Therefore, it is crucial to know the location of the sensor nodes. Location is required for data tagging, as well as, target detection, node tracking, etc. In addition, localization is essential for position-based routing algorithms which are powerful alter-natives to classical routing approaches in Mobile Ad Hoc Networks (MANET). Localization is a well studied topic in terrestrial sensor networks. Nevertheless, in Underwater Sensor Networks, localization is still challenging due to several reasons: i) unavailability of the GPS; ii) low bandwidth, long delay and high bit error rate of the acoustic links; iii) necessity of high amount of sensor nodes to cover the three dimensional region. The use of GPS is limited to surface nodes because the GPS signal does not propagate through the water. In sensor networking literature, several GPS-less (GPS-free) positioning schemes have been proposed however they usually have high overhead. The underwater sensors use acoustic links and the band-width of those links is low even for very short distances. Moreover, acoustic communications has high propagation delay and high bit error rate. In Underwater Sensor Networks, localization protocols are expected to avoid excessive overhead and establish localization with the least possible messages. This is also enforced by the limited battery life of the underwater sensor nodes and the difficulty of re-charging or replacing the batteries in an underwater application. Usually, an underwater application requires a large number of sensor nodes because the data rate of the acoustic links increases with decreasing distance and shorter ranges between nodes, means that more sensor nodes are needed to cover the three dimensional oceanographic zone. In addition, in a mobile Underwater Sensor Network localization should be repeated and stale location information should be cleared periodically. Considering all these challenges, it is essential to develop novel localization protocols tailored for mobile Underwater Sensor Networks. In this article, we introduce two distributed, scalable localization techniques; Dive and Rise Localization (DNRL) and Proxy Localization (PL). In DNRL, mobile beacons ascend and descend in the water to deliver their GPS driven coordinates. In PL, the al-ready localized nodes act like beacons likewise multi-stage localization which is a preliminary version of PL (Erol et al., 2008). Unlike the previous work, in PL the non-localized nodes use a different metric to choose the best possible proxies among the candidates which enhances the performance of the protocol. Here, we compare the performance of PL, DNRL and Large-Scale Localization (LSL). LSL is a technique from the literature (Zhou et al., 2007). We evaluate the performance of these schemes in terms of localization success, accuracy, overhead and energy consumption. Since we compare the performance of the localization techniques for a mobile Underwater Sensor Network, a realistic underwater mobility model is essential. Recently, the works of Caruso et al. (2008) and Erol et al. (2008) have applied the real ocean current behavior to Underwater Sensor Networks. We use the "Meandering Current Mobility with Surface Effect" (MCM-SE) model to compare the performance of the three localization schemes for a mobile underwater sensor network. The main aim of the article is to provide a comparison between recently proposed localization schemes for Underwater Sensor Networks. Based on the simulation results, we compare and analyze the performance of three recent methods that are developed for distributed localization in large-scale Underwater Sensor Networks in terms of localization ratio, accuracy, protocol overhead and energy consumption. [ABSTRACT FROM AUTHOR]