論文

The AKARI Project, one of clean slate approaches to design New Generation Network (NwGN), is proposing a six-layer network architecture called ZNA (Z Network Architecture). One of the features of ZNA is that the session layer (L5) is inserted between the application layer (L6) and the transport layer (L4) to provide the application layer with more sophisticated communication paths. The API and protocols of L5 have been proposed, however the behavior has not been confirmed by implementation. This paper designs in detail the protocols and APIs of L5, and verifies the feasibility by implementing a prototype of L5 in the user space before implementing it in the kernel. In this prototype, two new types of paths can be established: the bundled path and the spatiallyspliced path. In the former, bandwidth aggregation or fault tolerance is supported by using multiple L4 paths. In the latter, end-to-end communication via middle boxes is supported by splicing two L4 paths at the middle boxes after negotiation between the end node and the middle boxes. Copyright 2012 ACM.

To efficiently adapt to dynamically changing network environment such handover and link failure, cross-layer cooperation mechanisms are indispensable. This paper first introduces the following five requirements to cross-layer cooperation architecture through a discussion, (1) cooperation between a layer in a node and a layer (not necessary the same layer) in another node, (2) abstraction of control information, (3) definition of primitives in each layer, (4) availability in a wide area network, and (5) implementation. Although there are a lot of proposals of cross-layer cooperation architecture in the literature, none of them satisfy the five requirements. This paper designs an inter-node cross-layer cooperation architecture called CLINEX (Cross-Layer and Inter-Node information EXchange) so that it satisfies the five requirements. CLINEX is implemented in the user space and the kernel space of Linux. Two scenarios are tested in our test network composed of eight machines. One scenario is congestion detection at a remote node, which results in sending rate control at the end node. Another scenario is link failure detection at a remote node, which results in failover in an SCTP association. The measurement results show that the processing time of CLINEX is negligible in comparison with packet transmission latency. © 2013 IEEE.

To overcome MobileIPv6's natural centrality driven single point of failure and redundant routing due to central Home Agent in home network, Global HAHA introduced an overlay of scattered multiple HAs. Placements of secondary HAs were not addressed by major work in existence. This paper analyses placements in ISP's viewpoint at AS level. Freeman's closeness index and betweenness are used in categorizing examined AS level topology of 30, 000 nodes. 30, 000 different random communications were analyzed which substantiated that Freeman's closeness describes topological locations better. Results show that categories having highest, medium indices are tend to hold better performance only when combined with an HA of lowest category. Lowest category shows best performance with the highest category. © 2013 IEEE.

Sharing files over the world with higher access throughput and with lower storage cost is a growing demand for the applications that use large files. However, the existing file sharing systems do not realize these two conflicted requests and no research has been found. This paper clarifies the requirements of a global large file sharing system and defines the design goal consisting of three users perspectives: fast retrieving, user defined file availability, and owner-based file management, and one system operators perspective: flexibility in bytes placement. Content Espresso satisfies this goal by approaching with four techniques: three sections model, distributed chunk storage, forward error correction, and UDP retrieving. Content Espresso delivers large files to a client utilizing as much bandwidth as the client access link even servers are located far away from the client. © 2013 IEEE.

コンピュータ技術・ネットワーク技術の発展に伴い大容量コンテンツを利用する機会が増加している．しかしながら，大容量コンテンツをインターネットを介して高速かつ低ストレージコストで共有・保存することは難しい．筆者らは大容量コンテンツを対象として広域分散ストレージサーバに分散保存し，高速かつ低ストレージコストでインターネット規模のコンテンツ共有と可用性・信頼性のあるコンテンツ保存を実現するContent Espressoの設計と実装をしている．本論文では，Content Espressoの技術的要となるファイル保存・取得方式について述べる．本方式はネットワーク状態によらずファイル取得時間を予測でき，保存ファイルの信頼性・可用性を低冗長率で保証するファイル保存・取得方式である．本方式のアプローチは，1)前方誤り訂正(FEC: Forward Error Correction)符号を利用したファイル冗長化，2)データをチャンクに分割し，世界中のストレージサーバへ分散保存，3) UDPを利用したデータ取得の3点である．このアプローチに沿って本方式を設計・実装し，インターネット規模ネットワーク環境を想定した評価により，TCPを利用したファイル転送方式よりも，ファイル取得時間予想が容易であることを確認した．

We developed a fast handover mechanism using a high-speed image sensor for a high data rate ground-to-train free-space optical communication system in order to achieve a continuous communication between passengers and the Internet. Using the new handover mechanism, we were able to achieve handover time of 21 milliseconds in the best case, which is short enough to offer live streaming Internet services such as Skype, YouTube, and Internet Radio.

This paper investigates impacts of periodical burst packet loss on three typical streaming applications: YouTube, Internet Radio, and Skype, supposing a mobile network in a high-speed train. Although the characteristics of YouTube and Skype are well analyzed, the characteristics of Internet Radio and the characteristics of these applications under periodical burst packet loss have not been analyzed. First, this paper analyses how much these applications are tolerant to periodical burst packet loss. Second, this paper analyses correlation between application layer protocols and transport layer protocols. As a result, some correlations between transport layer protocols and application layer protocols were found. For example, in YouTube, if TCP segment retransmissions frequently occur before starting content download, a client tries to download the entire content with a single HTTP request while a client repeatedly downloads a chunk of the content with multiple HTTP requests under ordinary network condition. © 2014 IEEE.

Using "clean-slate approach" to redesign the Internet has attracted considerable attention. ZNA (Z Network Architecture) is one of clean-slate network architectures based on the layered model. The major features of ZNA are as follows: (1) introducing the session layer to provide the applications with sophisticated communication services, (2) employing inter-node cross-layer cooperation to adapt to the dynamically changing network conditions, (3) splitting the node identifier and the node locator for mobility, multi-homing, and heterogeneity of network layer protocols, (4) splitting the data plane and the control plane for high manageability, and (5) introducing a recursive layered model to support network virtualization. This paper focuses on the first three topics as well as the basic design of ZNA.

We propose a digital content viewing system which visualizes the polymorphic relationships among digital content interactively and uses it as a visual interface. The relationships are expressed by associating of two content and grouping and shown by directed graphs. On the graphs, nodes mean the content files or the groups, and edges mean correspondences or group members. The graphs are archived as relation catalogues separately from content files. We visualize the relations of content on a directed graph. To visualize the graphs, we use a method of locating nodes by horizontal directions and height. Nodes linked with a target node will be located on a horizontal circle and their height is decided by their relation catalogue. After deciding the node location, nodes and links are drawn using 3D graphics and animation. The visualization result called 'polymorphic view' is used as a visual interface for selecting content. Selecting content on the visualized graph will expand new links and new content nodes additionally. We can show another content by tracing graph links. It becomes easy to understand the relations between digital content, because we can change the direction and size of the directed graph. We developed a content viewer system (MoSaIC-II) using the polymorphic view as a visual interface and we can see the usability of the polymorphic view.

The literature lacks theoretical and practical evaluations of Distributed Mobility Management (DMM) schemes which aim to mitigate disadvantages of conventional centralized schemes such as a single point of failure, congestion in home network, and limited scalability. This paper selects Global HAHA, Migrating HA, and Distributed Mobile IP (DMIP) as representative host-based decentralized Internet Mobility Management (MM) schemes, and probes their control and data planes against Mobile IP (MIP) and MIP with Return Routability procedure (MIPRR), using real Autonomous System (AS) level topology data of the Internet. By using different combinations of MM Agent (MMA) placements, random sessions are investigated for host mobility in terms of the total number of hops and the AS load. Comparison of generic results shows that DMIP and Migrating HA outperform the legacy schemes. Further, DMIP overcomes consistency maintenance and scalability problems seen in Migrating HA, maintaining much similar performance, while achieving reliability eliminating an expensive controlling overhead as in MIPRR.

Nowadays each network management system (NMS) adopts different methods for collecting network information and different data structures. This makes NMS cooperation difficult, especially in multi-AS wide area network management. The current goal of the KANVAS (Knowledge base system in wide Area Networks with general Versatility, Availability and Scalability) project is to realize wide area network management by constructing a knowledge base system in networks. This paper discusses construction of a knowledge base system in a single AS (Autonomous System) towards the future wide area network management. This paper also shows the design and implementation of the knowledge base system that collects network configuration information from OSPF and SNMP, and stores it as instances of our proposed network ontology called Bonsai. Additionally, a prototype application that accesses the stored knowledge was implemented. The evaluation of basic performance and scalability of the system was carried out using a real AS topology and synthetic topologies and it was made sure that the prototype application works correctly. As a result, this paper shows that feasibility of the future wide area network management that adopts the knowledge base system.

As the quality of digital image is getting higher, the size of the images is getting larger. For example, an uncompressed 8bit HD color image file exceeds 6MB. Therefore, when we store and access large image files through network, high-speed storage I/O, CPU, and network are required. However, the state-of-the-art storage system is very expensive. Our dispersed networked storage system called Content Espresso consisting of off-the-shelf computers with Forward Error Correction and UDP-based delivery mechanisms realizes high-speed access to the stored large files. Our proof of concept prototype realized a series of HD uncompressed image files delivery from Tokyo to San Diego and accomplished a video playback at least 24 fps frame rate.

Distributed Mobility Management (DMM) defines Internet Protocol (IP) mobility which does not depend on centralized manipulation. DMM leads to the abatement of non-optimal routing, a single point of failure, and scalability problems appearing in centralized Mobility Management (MM). The fact that most DMM schemes are in the proposal phase and non-existence of a standardization, urge to investigate the proposed schemes thoroughly to confirm their capabilities and thereby, to determine the best candidate practice for DMM. This paper examines five novel DMM proposals discussed in the Internet Engineering Task Force (IETF) using router-level Internet Service Provider (ISP) topologies of Sprint (USA), Tiscali (Europe), Telstra (AUS), and Exodus (USA), as user mobility within an ISP network is considered the most realistic and recurrent user movement in the modern scope. Results reflect behavioral differences of schemes depending on the network. ISPs closer to the Internet core with high density of Point of Presences (PoPs) such as Sprint show poorer outcome when centralized anchors/controllers are employed while Proxy Mobile IP (PMIP) based enhancements offer higher reliability. In contrast, smaller ISPs that reside farther away from the Internet core yield better performance with SDN-Based and Address Delegation schemes. Although the PMIP-Based DMM schemes perform better during handover, their outturn is trivialized due to higher latency in the data plane. In contrast, the Address Delegation and SDN-Based schemes have excessive cost and latency in performing handover due to routing table updates, but perform better in data plane, suggesting that control/data plane split may best address the optimal routing.

There is a strong demand to enjoy broadband and stable Internet connectivity not only in office and the home but also in high-speed train. Several systems are providing high-speed train with Internet connectivity using various technologies such as leaky coaxial cable (LCX),Wi-Fi, and WiMAX. However, their actual throughputs are less than 2 Mbps. We developed a free-space optical (FSO) communication transceiver called LaserTrainComm2014 that achieves the throughput of 1 Gbps between the ground and a train. LaserTrainComm2014 employs a high-speed image sensor for coarse tracking and a quadrant photo-diode (QPD) for accurate tracking. Since the image captured by the high-speed image sensor has several types of noise, image processing is necessary to detect the beacon light of the other LaserTrainComm2014. As a result of field experiments in a vehicle test course, LaserTrainComm2014 achieves handover time of 21 milliseconds (ms) in the link layer at the speed of 60 km/h. Even if the network layer signaling takes time of 10 milliseconds, the total communication disruption time due to handover is short enough to provide passengers with Internet connectivity for live streaming Internet applications such as YouTube, Internet Radio, and Skype.

Yamata-no-Orochi is an authentication and authorization infrastructure across multiple service domains and provides Internet services with unified authentication and authorization mechanisms. In this paper, Yamata-no-Orochi is incorporated into a video distribution system to verify its general versatility as a multi-domain authentication and authorization infrastructure for Internet services. This paper also reduces the authorization time of Yamata-no-Orochi to fulfill the processing time constrains of the video distribution system. The evaluation results show that all the authentication and authorization processes work correctly and the performance of Yamata-no-Orochi is practical for the video distribution system.

Remote collaboration work via the Internet becomes common these days by appearance of web-based remote collaboration systems such as Google Apps, and SAGE2. This paper focuses on SAGE2 as a typical data intensive collaboration system, which shares information on large highresolution displays with other sites using web browser technologies. However, it takes long time to share large files between SAGE2 systems. In order to improve the file sharing performance of SAGE2, this paper proposes a file sharing mechanism using Content Espresso, a distributed storage system for global large file sharing, which can be applied to other web-based collaboration systems. We implemented and evaluated the proposed mechanism by emulating a global network environment. The result of the evaluation confirms that the proposed mechanism has better performance than the existing file sharing mechanism on SAGE2 when RTT between SAGE2 systems is longer than 10 ms.

DTN (Delay/Disruption Tolerant Networking) inserts a new layer between the application layer (L7) and the transport layer (L4) for reliable communication on long-delay, intermittent, or lossy inks. Such a new layer, however, should provide not only a single rich service but also various rich services. We are proposing Layer-5 (L5) inserted between L7 and L4 to provide multiple rich services. One of the rich services of L5 is the L5 temporally-spliced path (L5-TSP). It is an application-to-application path composed of multiple L4 paths, which may not exist at the same time, spliced in L5 in the nodes called splicers. This paper proposes L5-TSP Protocol (L5-TSPP) for reliable communication on lossy links. L5-TSPP places splicers at both ends of a lossy link. It uses TCP on reliable links while UDP on lossy links. It uses partial acknowledgement to enable end nodes to know progress of communication. L5-TSPP is implemented on Linux. It takes 61 msec for connection establishment in L5-TSPP while 530 msec in TCP at packet loss rate of 10 % on a lossy link. It takes 18 sec to send a 1GB file in L5-TSPP while 73 sec in TCP at packet loss rate of 1 %.

Recently, applications' demands for the network become more multifaceted. Highly functional application-to- application communication services such as bandwidth aggregation, fault tolerant communication, and delay/disruption tolerant networking (DTN) were developed independently in the network layer, the transport layer, and the application layer. As a result, protocol layering has become complicated. This paper proposes Layer-5 (L5) inserted between the application layer and the trans- port layer to separate communication policies and communication mechanisms to make protocol layering clearer. The transport layer (L4) provides end-to-end communication mechanisms such as reliable byte stream while L5 realizes communication policies such as bandwidth aggregation by combining the communication mechanisms in L4. This paper proposes five types of L5-paths as communication policies: (1) the L5 bundled path for bandwidth aggregation or fault tolerant communication, (2) the L5 spatially- spliced path for communication with middleboxes, (3) the L5 temporally-spliced path for DTN, (4) the L5 spliced- bundled path, and (5) the L5 bundled over spatially-spliced path. An application can select and use an appropriate L5- path depending on the network circumstances through a common API. A prototype of L5 is implemented in the Linux user space as a library to make deployment and maintenance easier. The evaluation shows that establishment time of L5- paths is short enough and performance of L5-paths is comparable or superior to existing technologies.

IEEE1588 (PTP: Precise Time Protocol), which aims for the sub-microsecond range of clock synchronization in a minimal network, is currently focused on highly precise clock synchronization. Boundary Clock (BC) and VLAN Priority are considered to be features that improve PTP accuracy. Although some papers addressed the accuracy, these evaluations were realized using different evaluation metrics and environments. We make them comparable to analyse the contribution of BC and VLAN Priority to precise clock synchronization. Our accuracy evaluation used PTP nodes available on the market and shows the results following the test cases defined in the ITU-T G.8261 Recommendation and IEEE1588 standard. The measurement results show that the accuracy was better than 200 nanoseconds with BC enabled, while it was about one microsecond with BC disabled. We observed that the accuracy was improved to approximately 850 nanoseconds when VLAN Priority was enabled.

With rapid growth of producing high-resolution digital contents such as Full HD, 4K, and 8K movies, the demand for low cost and high throughput sharing of content files is increasing at digital content productions. In order to meet this demand, we have proposed DRIP (Distributed chunks Retrieval and Integration Procedure), a storage and retrieval mechanism for large file sharing using forward error correction (FEC) and global dispersed storage. DRIP was confirmed that it contributes to low cost and high throughput sharing. This paper describes the design and implementation of Content Espresso, a distributed large file sharing system for digital content productions using DRIP, and presents performance evaluations. We set up experimental environment using 79 physical machines including 72 inexpensive storage servers, and evaluate file metadata access performance, file storage/retrieval performance, FEC block size, and system availability by emulating global environments. The results confirm that Content Espresso has capability to deal with 15,000 requests per second, achieves 1 Gbps for file storage, and achieves more than 3 Gbps for file retrieval. File storage and retrieval performance are not significantly affected by the network conditions. Thus, we conclude that Content Espresso is capable of a global scale file sharing system for digital content productions.