The First International Symposium on Computing and Networking (CANDAR 2013) was held in Matsuyama, Japan, from December 4^th to 6^th, 2013. This issue includes the extended version of 12 (twelve) papers that have been presented at the symposium. The first 4 (four) papers have been selected from invited keynote and tutorial talks of CANDAR 2013. Each of these papers has been thoroughly reviewed by experts. The final version, listed below, incorporates the reviewers' comments and suggestions.

抄録全体を表示

Novel optical network architectures are proposed for creating future network services. The first architecture is a centralized approach for higher energy efficiency; it yields a data center-centric metro/access optical aggregation network based on wavelength/time-slot multiplexing. Not only higher application layer functions but also all layer-3 or upper traffic are transferred through the simple metro/access optical aggregation network and switched in a huge centralized giant router at the data center. Its simple optical switching is 200 times more energy efficient and only one electrical router is needed, so power consumption of the network can be reduced ten or twenty fold compared to the existing Internet. The second is service mash-up by linked data through a network that uses broadband optical wire for the IoT era. All service contents, hardware, and software programs are defined as service parts. Optical wire interconnects some service parts and creates new mash-up services in the network. That creates deep network functionality in combination with network and processing functions. Both two approaches are quite simple and energy efficient in comparison to existing Internet, so they can be applicable future data center network architecture.

抄録全体を表示

Most researchers look for topics from the literature. But our research derived mostly from development, in turn driven by industrial projects or product testing. We spanned into the areas of cable TV networks, multi-hop cellular, Internet QoS, deep packet inspection, traffic forensics, embedded benchmarking, and software defined networking. Among them, our multi-hop cellular work was the first along this line and has a high impact on both academia and industry, with over 600 citations and standardizations in WLAN mesh (IEEE 802.11s), WiMAX (IEEE 802.16j), Bluetooth (IEEE 802.15.5), and 3GPP LTE-advanced. Side products from our research include a startup (L7 Networks Inc., in 2002), a test lab (Network Benchmarking Lab, NBL, since 2002), and a textbook “Computer Networks: An Open Source Approach” (McGraw-Hill, 2011). It is a perfect time to have my 20-year half-time report as we celebrate the 70th birthday of my Ph.D. thesis advisor, Prof. Mario Gerla. This report could serve as a reference for researchers in developing their own roadmap.

抄録全体を表示

The emergence of compute unified device architecture (CUDA), which has relieved application developers from having to understand complex graphics pipelines, has made the graphics processing unit (GPU) useful not only for graphics applications but also for general applications. In this paper, we present a cycle sharing system named GPU grid, which exploits idle GPU cycles to accelerate scientific applications. Our cycle sharing system implements a cooperative multitasking technique, which is useful for remotely executing a guest application on a donated host machine without causing a significant slowdown on the host. In addition, our system estimates whether a GPU is busy, partially idle, or fully idle, to accordingly maximize guest application throughput. Experimental results show that our system not only avoids frame rate degradation but also achieves a 91\% higher guest application throughput in comparison to a previous system that estimates GPU load by monitoring mouse and keyboard activities.

抄録全体を表示

Distance constrained labeling problems, e.g., L(p,q)-labeling and (p,q)-total labeling, are originally motivated by the frequency assignment. From the viewpoint of theory, the upper bounds on the labeling numbers and the time complexity of finding a minimum labeling are intensively and extensively studied. In this paper, we survey the distance constrained labeling problems from algorithmic aspects, that is, computational complexity, approximability, exact computation, and so on.

抄録全体を表示

The hierarchical dual-net (HDN) was introduced as a topology of interconnection networks for extremely large parallel computers. The HDN is constructed based on a symmetric product graph (base network). A k-level hierarchical dual-net, HDN(B, k, S), contains (2N0)2k /(2 ×􏰂Π􏰂ki=1 si) nodes, where S = {G′1, G′2, . . . , G′k}, G′i is a super-node and si = |G′i| is the number of nodes in the super-node at the level i for 1 ≤ i ≤ k, and N0 is the number of nodes in the base network B. The node degree of HDN(B, k, S) is d0 + k, where d0 is the node degree of the base network. The HDN is node and edge symmetric and can contain huge number of nodes with small node-degree and short diameter. Disjoint-path routing is a fundamental and critical issue for the performance of an interconnection network. In this paper, we propose an efficient algorithm for finding disjoint-paths on an HDN and give the performance simulation results.

抄録全体を表示

This paper studies task scheduling algorithms which schedule a set of tasks on multiple cores so that the total scheduling length is minimized. Most of the algorithms developed in the past assume that a task is executed on a single core. Unlike the previous algorithms, the algorithms studied in this paper allow a task to be executed on multiple cores. This paper proposes six algorithms. All of the six algorithms are based on list scheduling, but the strategy for priority assignment is different. In our experiments, the six algorithms as well as an integer linear programming method are evaluated.

抄録全体を表示

CUBIC TPC is a congestion control algorithm for TCP. It is the current default TCP algorithm in Linux. Because many Internet servers, such as web servers, are running on Linux operating system, keeping throughput obtained with this TCP enough is quite important. Then, many performance studies have been published. However, most of these studies have been based on network simulators. Thus, evaluations using an actual TCP implementation and actual network elements are important in addition to these existing studies. In this paper, we focus on RTT (round trip time) fairness on CUBIC TCP, which is performance fairness among CUBIC TCP connections with different network delay times. Firstly, we present RTT fairness evaluation using actual TCP implementations and actual network elements and show that the fairness is not enough. Secondly, we discuss the cause of the unfairness based on CUBIC TCP behaviors. Thirdly, we propose a method for improving RTT fairness of CUBIC TCP. Unlike an existing work, the proposed method is not based on heuristic optimization. Finally, we present evaluation results and demonstrate that the proposed method provides better fairness than original CUBIC TCP implementation.

抄録全体を表示

Many-core processors provide the raw computation power required by modern high-performance multimedia and signal processing workloads. The conversion of this computation power into execution performance is often constrained by the overheads of communication between concurrent tasks. This paper presents Pronto, a low overhead message passing system which simplifies the semantics of data movement between communicating tasks by performing buffer management, message synchronization and address translation directly in hardware. The integration of these functions into hardware results in transfer latencies upto 30% shorter than state of the art MPI derivatives. The overheads for communication with Pronto in an 18-core processor array are under 5% for 64-word burst transfers, and less than 0.5% of total execution time using workloads such as the JPEG decoder and FIR filter. Furthermore, this paper also studies the effect of task mapping and interconnect traffic on the predictability of data block arrival times, and provides insight on where interconnect contention can be tolerated.

抄録全体を表示

In this paper, we propose a basic technique to minimize the computational time in executing the infinite-stage dynamic programming (DP) on a GPU. The infinite-stage DP involves computations to probe whether a value function gets sufficiently close to the optimal one. Such computations for probing convergence become obvious when an infinite-stage DP is executed on a GPU, since those computations are not necessary for finite-stage DPs, and hide behind loops for updating state values when a DP is executed on a CPU. The heart of the proposed technique is to suppress those computations for probing by thinning out them. By the proposed technique, differences between state values before and after being updated are periodically transferred to the main memory, then are checked to probe convergence. This intermittent probing makes contrast to ordinary methods in which computations for probing are processed every time. The technique also proposes a formulation to determine optimal periods for probing based on simple statistics given by preliminary experiments. The effectiveness of the proposed technique is examined on two problems; the one is a kind of the animat problem in which an agent moves around in a maze to collect foods, and the other is the mountain-car problem in which a powerless car on a slope struggles to pass over a higher peak. Computational results display that a method with the proposed technique decreases computational times for both problems compared to methods in which computations for probing convergence are processed every time, and the degree of decreasing seems remarkable when the state space is larger.

抄録全体を表示

In wireless ad hoc networks, bridges and articulation nodes are critical elements that, in case of failure, render the network disconnected. Owing to their relevance, a number of works try to extend the life span of these elements. Nevertheless, in critical situations, such as the unavailability of a critical link, ways to reestablish the communication, even if for short periods of time, can be of importance in a number of urgent tasks. In this context, this work explores the concept of Cooperative Communication (CC) to monitor critical nodes and links and recover network connectivity in case of disruption. Unlike other works that perform exhaustive search to locate suitable CC-links that require global topology information, the proposed scheme identifies critical nodes and links based solely on local information. Compared to other prominent works, the proposed solution was able to reduce the computing cost to create CC-links in ≈ 67 times in the evaluated scenarios while persisting a lower message overhead.

抄録全体を表示

Internet of Things (IoT) has attracted much attention today. As the key component of IoT, the Radio Frequency Identification (RFID) network has been growing in usage. An RFID network is composed of a backend system (including the database), one or more readers, and several RFID tags. Most of the RFIT tags cannot transmit signals actively, except the active RFID tags. The active RFID tags may gather information from sensors and store the information for later delivery to the reader. To extend the coverage of the RFID network, some RFID multi-hop relay systems (using active RFID tag as relay tags) have been proposed. They are all designed in ”Reader Talks First (RTF)” mode. That means multi-hop connections may be built up only when the reader queries and the tags response. On the contrary, in "Tag Talks First (TTF)" mode, an active tag cannot report anything to the system if it cannot find an available reader in its sense area. It may occur when the reader is located out of the sense area of the tag, or the reader is out of service (but the tag still needs to find out another reader to join in the RFID network). If these active tags are connection-lost from the system, some emergence messages may be lost. In this paper, we propose two multi-hop relay algorithms used in TTF mode, which are designed for Free2move and 802.15.4 standard respectively. Simulation results show that with the help of the proposed algorithms, the loss performance in TTF mode is improved a lot. Besides, the energy performance and delay performance is also improved compared to not using multi-hop relay algorithms in TTF mode, but use periodical data collection in RTF modes instead.

抄録全体を表示

Image analysis is crucial to medical and biological applications. Recent advances in imaging technology have led to the demand for processing and visualizing a large amount of three-dimensional (3D) biomedical images. In addition, cloud computing has become popular for managing big data. Unfortunately, conventional image-processing systems either lack cloud computing services or advanced 3D processing abilities. In this paper, we present a novel cloud-based system for sharing, processing, and visualizing 3D biomedical images. Our system employs a standard web browser as a client interface that interactively communicates with high-performance servers. Thus, an inexpensive tablet PC without an advanced graphics processing unit (GPU) can be used for 3D image processing and visualization. Our system provides the sharing of limited software and hardware resources, and it allows for effective collaboration between researchers. We demonstrate the applicability and functionality of the system by examining typical case studies on biomedical images. We also examine the performance of our system numerically.

抄録全体を表示