Crosstalk in optical switch is an intrinsic drawback of optical networks, and avoiding crosstalk is important for making optical network work properly. Horizontal expansion and vertical stacking are two basic techniques for creating nonblocking multistage interconnection networks (MINs). Rearrangeable (nonblocking) optical MINs are feasible since they have lower complexity than their strictly nonblocking counterparts. In this paper, we study the crosstalk-free permutations in rearrangeable optical MINs built on a combination of horizontal expansion and vertical stacking of banyan networks, and provide a scheme for realizing crosstalk-free permutations in this kind of optical MINs. The basic idea of this scheme is to first decompose a permutation into multiple partial permutations by using Euler Split technique, then route and realize each of these partial permutations crosstalk-free in one plane (stacked copy) of a MIN based on both the Euler Split technique and self-routing property of a banyan network. The tradeoff between the overall time complexity and hardware cost of this class of MINs is also explored in this paper.

Effects of the element radiation pattern on characteristics of the spherical array antenna composed of turnstile elements is discussed. It is found by numerical calculations that scanning characteristics of the directivity and the axial ratio do not so much depend on element radiation patterns.

Conditional random fields (CRFs) have been successfully applied to various applications of predicting and labeling structured data, such as natural language tagging & parsing, image segmentation & object recognition, and protein secondary structure prediction. The key advantages of CRFs are the ability to encode a variety of overlapping, non-independent features from empirical data as well as the capability of reaching the global normalization and optimization. However, estimating parameters for CRFs is very time-consuming due to an intensive forward-backward computation needed to estimate the likelihood function and its gradient during training. This paper presents a high-performance training of CRFs on massively parallel processing systems that allows us to handle huge datasets with hundreds of thousand data sequences and millions of features. We performed the experiments on an important natural language processing task (text chunking) on large-scale corpora and achieved significant results in terms of both the reduction of computational time and the improvement of prediction accuracy.

The multi-log2N network architecture is attractive for constructing optical switches, and the related routing algorithms are critical for the operation and efficiency of such switches. Although several routing algorithms have been proposed for multi-log2N networks, a full performance comparison among them has not been published up to now. Thus, we rectify this omission by providing such a comparison in terms of blocking probability, time complexity, hardware cost and load balancing capability. Notice that the load balance is important for reducing the peak power requirement of a switch, so we also propose in this paper a new routing algorithm for optical multi-log2N networks to achieve a better load balance.

Secure sensor localization is a prerequisite for many sensor networks to retrieve trustworthy data. However, most of existing node positioning systems were studied in trust environment and are therefore vulnerable to malicious attacks. In this work, we develop a robust node positioning mechanism(ROPM) to protect localization techniques from position attacks. Instead of introducing countermeasures for every possible internal or external attack, our approach aims at making node positioning system attack-tolerant by removing malicious beacons. We defeat internal attackers and external attackers by applying different strategies, which not only achieves robustness to attacks but also dramatically reduces the computation overhead. Finally, we provide detailed theoretical analysis and simulations to evaluate the proposed technique.

Interconnection networks usually suffer from Little's Law: low cost implies low performance and high performance is obtained high cost. However, hierarchical interconnection networks provide high performance at low cost by exploring the locality that exists in communication patterns of massively parallel computers. In this paper, we propose a new hierarchical interconnection network, called Hierarchical Torus Network (HTN). This network reduces the number of vertical links in 3D stacked implementation while maintaining good network features. This paper addresses the architectural details of the HTN, and explores aspects such as the network diameter, average distance, bisection width, peak number of vertical links, and VLSI layout area of the HTN as well as for several commonly used networks for parallel computers. It is shown that the HTN possesses several attractive features including small diameter, small average distance, small number of wires, a particularly small number of vertical links, and economic layout area.

Arrayed waveguide grating (AWG) is a promising technology for constructing high-speed large-capacity WDM switches, because it can switch fast, is scalable to large size and consumes little power. To take the full advantage of high-speed AWG, the routing control of a massive AWG-based switch should be as simple as possible. In this paper, we focus on the self-routing design of AWG-based switches with O(1) constant routing complexity and propose a novel construction of self-routing AWG switches that can guarantee the attractive nonblocking property for both the wavelength-to-wavelength and wavelength-to-fiber request models. We also fully analyze the proposed design in terms of its blocking property, hardware cost and crosstalk performance and compare it against traditional designs. It is expected that the proposed construction will be useful for the design and all-optical implementation of future ultra high-speed optical packet/burst switches.

The ray-tracing algorithm can synthesize very realistic images. However, the ray tracing is very time consuming. To solve this problem, a load balancing strategy using temporal coherence between images in an animation is presented for balancing computational loads among processing elements of a parallel processng system. Our parallel processing model is based on a space subdivision method for the ray-tracing algorithm. A subdivided object space is distributed among processing elements of the parallel system. To clarify the effectiveness of the load balancing strategy, we examine the system performance by computer simulation.

Vertical stacking is a novel technique for building switching networks, and packing multiple compatible connections together is an effective strategy to reduce network hardware cost. In this paper, we study the crosstalk-free permutation capability of an optical switching network built on the vertical stacking of optical banyan networks to which packing strategy is applied. We first look into the nonblocking condition of this optical switching network. We then study the crosstalk-free permutation in this network by decomposing a permutation evenly into multiple crosstalk-free partial permutations (CFPPs) and realizing each CFPP in a stacked plane of the network such that a crosstalk-free permutation can be performed in a single pass. We present a rigorous proof of CFPP decomposability of a permutation and also a complete algorithm for CFPP decomposition. The possibility of a tradeoff between the number of passes and the number of planes required for realizing a crosstalk-free permutation in this network is also explored in this paper.

We have constructed a parallelizing compiler for Standard ML (SML) based upon algorithmic skeletons. We present an implementation of a Parallel Molecular Dynamics (PMD) simulation in order to compare our functional approach with a traditional imperative approach. Although we present performance data, the principal benefits from our approach are in the modularity of the code and the ease of programming. Extant FORTRAN90 code for an O(N 2) algorithm is translated, firstly into imperative SML and then into purely functional SML which is then parallelized. The ease of programming and the performance of the FORTRAN90 and SML code are compared. Modest parallel performance is obtained from the parallel SML but with a much slower sequential execution time compared to the FORTRAN90. We then improve the implementation with a ring topology implementation which gives much closer performance to the FORTRAN90 implementation.

The evolution of VLSI chips towards larger die size, smaller feature size and faster clock speed makes the clock distribution an increasingly important issue. In this paper, we propose a new clock distribution network (CDN), namely Variant X-Tree, based on the idea of X-Architecture proposed recently for efficient wiring within VLSI chips. The Variant X-Tree CDN keeps the nice properties of equal-clock-path and symmetric structure of the typical H-Tree CDN, but results in both a lower maximal clock delay and a lower clock skew than its H-Tree counterpart, as verified by an extensive simulation study that incorporates simultaneously the effects of process variations and on-chip inductance. We also propose a closed-form statistical models for evaluating the skew and delay of the Variant X-Tree CDN. The comparison between the theoretical results and the simulation results indicates that the proposed statistical models can be used to efficiently and rapidly evaluate the performance of the variant X-Tree CDNs.

This paper proposes new algorithms for diagnosing (detection, identification and location) baseline multistage interconnection networks (MIN) as one of the basic units in a massively parallel system. This is accomplished in the presence of single and multiple faults under a new fault model. This model referred to as the geometric fault model, considers defective crossing connections which are located between adjacent stages, internally to the MIN (therefore, a fault corresponds to a physical bridge fault between two connections). It is shown that this type of fault affects the correct geometry of the network, thus requiring a different testing approach than previous methods. Initially, an algorithm which detects the presence of bridge faults (both in the single and multiple fault cases), is presented. For a single bridge fault, the proposed algorithm locates the fault except in an unique pathological case under which it is logically impossible to differentiate between two equivalent locations of the fault (however, the switching element affected by this fault is uniquely located). The proposed algorithm requires log2 N test vectors to diagnose the MIN as fault free (where N is the number of input lines to the MIN). For fully diagnosing a single bridge fault, this algorithm requires at most 2 log2 N tests and terminates when multiple bridge faults are detected. Subsequently, an algorithm which locates all bridge faults is given. The number of required test vectors is O(N). Fault location of each bridge fault is accomplished in terms of the two lines in the bridge and the numbers of the stages between which it occurs. Illustrative examples are given.

Recently, a promising packet forwarding architecture COPE was proposed to essentially improve the throughput of multihop wireless networks, where each network node can intelligently encode multiple packets together and forward them in a single transmission. However, COPE is still in its infancy and has the following limitations: (1) COPE adopts the FIFO packet scheduling and thus does not provide different priorities for different types of packets. (2) COPE simply classifies all packets destined to the same nexthop into small-size or large-size virtual queues and examines only the head packet of each virtual queue to find coding solutions. Such a queueing structure will lose some potential coding opportunities, because among packets destined to the same nexthop at most two packets (the head packets of small-size and large-size queues) will be examined in the coding process, regardless of the number of flows. (3) The coding algorithm adopted in COPE is fast but cannot always find good solutions. In order to address the above limitations, in this paper we first present a new queueing structure for COPE, which can provide more potential coding opportunities, and then propose a new packet scheduling algorithm for this queueing structure to assign different priorities to different types of packets. Finally, we propose an efficient coding algorithm to find appropriate packets for coding. Simulation results demonstrate that this new COPE architecture can further greatly improve the node transmission efficiency.

We propose an ant-based algorithm to improve the alternate routing scheme for dynamic Routing and Wavelength Assignment (RWA) in all-optical wavelength-division- multiplexing (WDM) networks. In our algorithm, we adopt a novel twin routing table structure that comprises both a P-route table for connection setup and a pheromone table for ants' foraging. The P-route table contains P alternate routes between a source-destination pair, which are dynamically updated by ant-based mobile agents based on current network congestion information. Extensive simulation results upon the ns-2 network simulator indicate that by keeping a suitable number of ants in a network to proactively and continually update the twin routing tables in the network, our new ant-based alternate routing algorithm can result in a small setup time and achieve a significantly lower blocking probability than the promising alternate shortest-path (ASP) algorithm and the fixed-paths least congestion (FPLC) algorithm for dynamic RWA even with a small value of P.

A combination of horizontal expansion and vertical stacking of optical Banyan (HVOB) is the general architecture for building Banyan-based optical cross-connects (OXCs), and the intrinsic crosstalk problem of optical signals is a major constraint in designing OXCs. In this paper, we analyze the blocking behavior of HVOB networks and develop the lower bound on blocking probability of a HVOB network that is free of first-order crosstalk in switching elements. The proposed lower-bound is significant because it provides network designers an effective tool to estimate the minimum blocking probability they can expect from a HVOB architecture regardless what kind of routing strategy to be adopted. Our lower bound can accurately depict the overall blocking behavior in terms of the minimum blocking probability in a HVOB network, as verified by extensive simulation based on a network simulator with both random routing and packing routing strategies. Surprisingly, the simulated and theoretical results show that our lower bound can be used to efficiently estimate the blocking probability of HVOB networks applying packing strategy. Thus, our analytical model can guide network designers to find the tradeoff among the number of planes (stacked copies), the number of SEs, the number of stages and blocking probability in a HVOB network applying packing strategy.

Ad-Hoc networks are collections of mobile nodes communicating using wireless media without any fixed infrastructure. Minimal configuration and quick deployment make Ad-Hoc networks suitable for emergency situations like natural disasters or military conflicts. The current Ad-Hoc networks can only support either high mobility or high transmission rate at a time because they employ static approaches in their routing schemes. However, due to the continuous expansion of the Ad-Hoc network size, node-mobility and transmission rate, the development of new adaptive and dynamic routing schemes has become crucial. In this paper we propose a new routing scheme to support high transmission rates and high node-mobility simultaneously in a big Ad-Hoc network, by combining a new proposed packet-pheromone-based approach with the Hint Based Probabilistic Protocol (HBPP) for congestion avoidance with dynamic path selection in packet forwarding process. Because of using the available feedback information, the proposed algorithm does not introduce any additional overhead. The extensive simulation-based analysis conducted in this paper indicates that the proposed algorithm offers small packet-latency and achieves a significantly higher delivery probability in comparison with the available Hint-Based Probabilistic Protocol (HBPP).

With the wide expansion of voice services over the IP networks (VoIP), the volume of this delay sensitive traffic is steadily growing. The current packet schedulers for IP networks meet the delay constraint of VoIP traffic by simply assigning its packets the highest priority. This technique is acceptable as long as the amount of VoIP traffic is relatively very small compared to other non-voice traffic. With the notable expansion of VoIP applications, however, the current packet schedulers will significantly sacrifice the fairness deserved by the non-voice traffic. In this paper, we extend the conventional Deficit Round-Robin (DRR) scheduler by including a packet classifier, a Token Bucket and a resource reservation scheme and propose an integrated packet scheduler architecture for the growing VoIP traffic. We demonstrate through both theoretical analysis and extensive simulation that the new architecture makes it possible for us to significantly improve the fairness to non-voice traffic while still meeting the tight delay requirement of VoIP applications.

A new parallel algorithm for the numerical integration of ordinary differential equations is proposed. The algorithm is based on the predictor and corrector formulae and their iterative use. An equation which describes the error propagation of the proposed algorithm is derived and discussed. The reliability and convergence properties of the solutions are studied by performing the numerical experiments on a conventional computer system for some specific examples. It is found that the proposed algorithm is indeed a practical one for a concurrentt system such as a multiprocessor and a vector processor, and offers much greater computing speed than the existing algorithms.

Recently, network coding has been applied to the loss recovery of reliable multicast in wireless networks, where multiple lost packets are XOR-ed together as one packet and forwarded via single retransmission, resulting in a significant reduction of bandwidth consumption. In this paper, we first prove that maximizing the number of lost packets for XOR-ing, which is the key part of the available network coding-based reliable multicast schemes, is actually a complex NP-complete problem. To address this limitation, we then propose an efficient heuristic algorithm for finding an approximately optimal solution of this optimization problem. Furthermore, we show that the packet coding principle of maximizing the number of lost packets for XOR-ing sometimes cannot fully exploit the potential coding opportunities, and we then further propose new heuristic-based schemes with a new coding principle. Simulation results demonstrate that the heuristic-based schemes have very low computational complexity and can achieve almost the same transmission efficiency as the current coding-based high-complexity schemes. Furthermore, the heuristic-based schemes with the new coding principle not only have very low complexity, but also slightly outperform the current high-complexity ones.

Cross-document personal name resolution is the process of identifying whether or not a common personal name mentioned in different documents refers to the same individual. Most previous approaches usually rely on lexical matching such as the occurrence of common words surrounding the entity name to measure the similarity between documents, and then clusters the documents according to their referents. In spite of certain successes, measuring similarity based on lexical comparison sometimes ignores important linguistic phenomena at the semantic level such as synonym or paraphrase. This paper presents a semantics-based approach to the resolution of personal name crossover documents that can make the most of both lexical evidences and semantic clues. In our method, the similarity values between documents are determined by estimating the semantic relatedness between words. Further, the semantic labels attached to sentences allow us to highlight the common personal facts that are potentially available among documents. An evaluation on three web datasets demonstrates that our method achieves the better performance than the previous work.