Naoki Oguchi

Reconfigurable TCP: An Architecture for Enhanced Communication Performance in Mobile Cloud Services

Commercial cloud services are growing rapidly, and system architecture that can process large volumes of large-scale data for mobile devices and sensing devices is expected to emerge in the future. Mobile devices will serve as both client terminals and virtual machines in such architecture. However, end-to-end TCP sessions are affected by frequent changes in wireless link quality and movement of virtual machines in the cloud data center. In this paper, we propose a method to improve communication performance by dynamically reselecting and retuning congestion control algorithms and parameters according to changes in link quality and attributes of applications.

RISP: address resolution protocol in network layer

Conventionally, the network layer (such as IP) to link layer (such as Ethernet and ATM) address translation is done in the link layer. This approach requires link technology-specific solutions such as ARP for Ethernet, and ATMARP or NHRP for ATM. By moving the address resolution function from link layer to network layer, this paper proposes a unified address resolution scheme applicable to current and future network technologies. For NBMA (non-broadcast multi-access) networks, this scheme achieves better implementation efficiency because the conventional scheme would require extra table lookups when the link layer has to consult with network-layer routing function to forward a request that cannot be resolved locally.

Estimating available bandwidth in mobile networks by correlation coefficient

In recent years, as mobile devices increase, the demands for measuring the available bandwidth in a wireless network increase more and more to provide high quality network services. The available bandwidth is one of the most important parameters to determine the sending rate in WAN optimization. Pathload [2] is a technique to measure the available bandwidth by detecting the point where the receiving rate becomes lower than the sending rate of test packets. However, Pathload misdetects this point in mobile networks with packet losses caused by fading, interference, and so on. So, we propose a method to measure the available bandwidth by using correlation coefficients between the sending rate and packet loss rate, and then evaluate the accuracy of the proposed method.

Dynamic Communication Protocol for Quick Response in Interactive Communication

Recently, many commercial cloud computing services have rapidly evolved, and it is expected that users will increasingly access these cloud services using mobile devices or thin client devices with response-sensitive applications such as remote desktop. For such communications, the transmission control protocol (TCP) is typically used between such devices and virtual machines in cloud data centers. However, frequent quality changes in the wireless link or movements of virtual machines influence end-to-end TCP sessions, and the communication response may be degraded. In this paper, we propose a method to improve the response in interactive communication by dynamically changing congestion control algorithms based on changes in the link quality. We then show the effectiveness of the proposed method.

Virtual data planes for easy creation and operation of end-to-end virtual networks

Today, it seems more and more required to create network slices dynamically and on-demand due to the emergence of cloud services, 5G wireless and IoT technology. It becomes possible to create virtual networks over physical networks easily by using such technology as Software Defined Networking (SDN) and Network Functions Virtualization (NFV). However, when creating end-to-end network slices over some infrastructures that are operated by different types of providers or companies, it is hard for system administrators who are not network experts to design such network slices, because those administrators have to understand virtual networks such as VPN and VXLAN for layer3 or layer2 tunnels between infrastructures. Therefore, we propose an architecture to create and operate network slices without awareness of complex virtual networks that are essential for inter- or intra-infrastructures, by creating abstraction layer (virtual network object), which intermediate between a logical network designed by a user and a virtual network overlaid on infrastructures.

Seamless Network Service Orchestration across On-Premises and Cloud Infrastructures

The emergence of 5G wireless and IoT (Internet of Things) technologies, as well as the continuous evolution of cloud computing is driving the need for easy and on-demand creation of end-to-end network services. We have developed an integrated system called the end-to-end network service orchestrator, which realizes seamless service orchestration across multiple on-premises and cloud infrastructures. By leveraging the virtual network object based service abstraction, a user can design and operate network services independent of the underlying infrastructure. We show a demonstration of service migration using this system across multiple cloud systems.

Application and evaluation of distributed WAN optimization technique in heterogeneous networks

WAN optimization becomes a well-known technique to improve transmission rate in file transfer applications and interactive applications served in international cloud services. WAN optimization technique can improve transmission speed by using high performance protocols instead of TCP between both ends of wide area network, which has a large round trip time and a high packet loss rate. There are some kinds of high performance protocols, and their effectiveness in improving communication are different from each other based on the characteristics of the networks and applications. When WAN optimization technique is only used at both ends of a heterogeneous network containing networks such as wireless networks, international leased lines, etc. (e.g., access from a mobile device to a foreign cloud data center), improvement in communication is limited because only one high performance protocol is being applied. Therefore, we propose “distributed WAN optimization technique” that dynamically changes high performance protocols and sections to adapt those protocols based on network characteristics and applications.

Performance Evaluation of Reconfigurable TCP Communication on Wi-Fi Network Using GE Channel Model

Recently, many commercial cloud-computing services have rapidly evolved to process and back up large volumes of data. With wireless broadband networks also gaining popularity, users increasingly access cloud services using mobile devices and wearable devices for interactive communication, such as remote desktops and augmented reality (AR) applications. In many cases, interactive applications use the TCP protocol between a server and a mobile device. However, end-to-end TCP sessions are easily influenced by packet losses in wireless networks and long delays on international lines, which may degrade communication quality. To cope with these problems, we propose a dynamic TCP communication method that improves the response for interactive communication by dynamically changing congestion control algorithms in accordance with changes in network quality and application characteristics. In this paper, we implement the proposed method on the NS-3 network simulator and evaluate it by using the GE (Gilbert-Eliott) channel model, assuming that the method is applied to a Wi-Fi network. We find that our method improves response time for interactive communication with a remote desktop to the level where users will not feel uncomfortable.

Implementation and Analysis of Scalable and Flexible Node for Large-scale Networks

It is necessary to improve the performance of packet forwarding nodes that form part of large-scale and functionally advanced networks to ensure that they function easily and effectively. In this paper, we propose a separation method, in which functions in the network node are divided into forwarding functions and control functions, and each set of functions is executed on different hardware (forwarder and controller). With this method, we configure resources in the forwarder as the controllers own virtual resources. By doing so, the software, which runs on the controller, is able to treat the forwarders resources as its own. We evaluate the effect of this method by applying it to a router, and clarify the performance improvement by the proposed method. And we select a stateful packet filtering as another test target, which executes the complex processing in packet forwarding. We then examine the field where the pro-posed method is effective