To realize an efficient DTN (Delay/Disruption-Tolerant Networking) routing, it is required to quickly deliver the message from the source node to the destination node as well as to quickly delete disused message replicas from the network. Epidemic routing, which indefinitely forwards message replicas to all encountered nodes, realizes the near-optimal message delivery delay when a limited number of messages are transferred. However, its performance is significantly degraded when a number of messages are transferred simultaneously. In our previous work, we have proposed a simple but effective extension to epidemic routing called restrained epidemic routing, which intentionally suppresses message forwardings at the later stage of epidemic-style message dissemination. In this paper, we analyze the characteristics of restrained epidemic routing when the contact relation between nodes is given by a general contact model such as complex networks. Specifically, we describe the dynamics of restrained epidemic routing on a complex network with a given degree distribution as differential equations using the degree-based mean field approximation.

DTN routing aims to realize message delivery from a node (source node) in the network to another specific node (destination node) without using dedicated communication infrastructure. Depending on the mobility of the source and the destination nodes, DTN routing is classified into four classes: mobile-to-mobile, fixed-to-mobile, mobile-to-fixed, and fixed-to-fixed. Most of conventional DTN routing studies have been focusing on mobile-to-mobile DTN routing. A large number of mathematical analyses and performance evaluations of mobile-to-mobile DTN routing have been performed, but the characteristics of other classes - in particular, mobile-to-fixed and fixed-to-fixed DTN routing - have not been well clarified. In this paper, as a mobile-to-fixed and fixed-to-fixed DTN routing, we focus on a position-based single-copy DTN routing. We build an analytical framework for those DTN routing mechanisms. Namely, we describe the average behavior of message delivery between the source node and the destination node. We then derive the average message delivery delay in the position-based single-copy DTN routing. Through numerical examples, we quantitatively compare the performances of a multi-copy mobile-to-mobile DTN routing and the position-based and single-copy mobile-to-fixed DTN routing.

PTP (Precision Time Protocol) is a network protocol for achieving more precise time synchronization among networked devices than the conventional NTP (Network Time Protocol). Although hardware-assisted timestamp during PTP message exchanges has been commonly used for realizing a high-precision time synchronization, software-only implementations are also studied in the literature. A software-only implementation must cope with jitter and noise observed in timestamps in PTP messages so that undesirable effects such as clock fluctuation and/or desynchronization can be mitigated. A software-only implementation of the PTP protocol called PTPd includes a clock adjustment mechanism (clock servo), which utilizes two types of low-pass filters and a PI (Proportional Integral) controller in classical control theory. In this paper, we build a discrete model of a networked system utilizing the PTP protocol and the clock adjustment mechanism at the slave device. We also analyze the stability of the entire system. Through numerical examples, we investigate the effect of control parameters of the clock adjustment mechanism on both steady-state and transient-state characteristics.

Advanced Driver-Assistance System (ADAS) technology has been developed in recent years for realizing an automotive network based on Ethernet and serial-bus based communication such as Control Area Network (CAN). EthernetAVB (Audio/Video Bridging) is a promising technique for realizing highly-reliable and low-latency Ethernet communication. Transport protocols for realizing low-latency communication (e.g., low-latency TCP) in data center networks have also been studied extensively. However, these communication standards are implemented in limited network devices and operating systems, making it difficult to construct an experiment environment for complex automotive networks comprising many Electronic Control Units (ECUs). In this paper, we propose a network emulator called ATINET (AuTomotive IP NETwork emulator), which supports EthernetAVB (802.1Qav) in layer-2 and ECN (RFC 3168) in layer-3 using network virtualization techniques in the Linux operating system.

In this paper, we propose a delivery control method of floating contents called PFCS (Proportional control for Floating Content Sharing) and investigate its properties through stability analysis. By intentionally limiting the coverage and the lifetime of epidemic broadcasting, floating contents can be shared among mobile nodes without dedicated infrastructure. Information sharing with floating contents is realized by (1) embedding the usable area (i.e., anchor zone) and the lifetime (i.e., TTL (Time-to-Live)) in a message, (2) forwarding the message among mobile nodes only in its anchor zone, (3) deleting the message if it expires, and (4) deleting the message, if necessary, once the mobile node carrying the message leaves the anchor zone. Hence, if message forwarding among mobile nodes is discontinued or some messages are lost due to buffer overflow of a mobile node, floating contents may be vanished. Such a limitation becomes more problematic when the anchor zone accommodates a number of floating contents. In this paper, we therefore propose a delivery control method of floating contents called PFCS, which controls the message possession ratio (i.e., the fraction of mobile nodes carrying a message in the anchor zone). We also perform stability analysis of PFCS to investigate fundamental properties of PFCS. Through numerical examples and simulation results, we demonstrate the effectiveness of PFCS as well as the validity of our approximate analysis.

In this paper, we derive the average message delivery of epidemic routing with broadcasting ACKs (ACKnowledgements) in DTN (Delay/Disruption-Tolerant Networking). The epidemic routing achieves near-optimal performance in terms of the message delivery delay when there exists only a single message in the network. However, if there exist multiple messages, epidemic routing generates excessive amount of message copies, resulting in poor performance. One of the promising techniques to alleviate the drawbacks of epidemic routing is broadcasting ACKs, which propagates information on the successful delivery of the message to all other nodes to eliminate unnecessary copies to avoid the waste of network bandwidth. In the literature, the performance of epidemic routing with broadcasting ACKs for a single message has been analyzed. However, to the best of our knowledge, the performance of epidemic routing with broadcasting ACKs under multiple concurrent message routings has not been well understood. In this paper, utilizing the Markov model of epidemic routing with broadcasting ACKs, we derive the average message delivery delay of epidemic routing with multiple messages, each of which competes for the network bandwidth.

In this paper, we derive the average and the distribution of message delivery delays in a geographic DTN routing with multiple mobile agents, whose mobility patterns are given by random walks on a graph and message routing algorithm is FIFO (First-In First-Out) algorithm. A geographic DTN routing aims at realization of message delivery among multiple (generally, geographically-dispersed) geographic locations on a field without necessity of specific communication infrastructure by utilizing mobility of mobile agents. We model the behaviors of mobile agents as multiple random walks on a graph. In this paper, two types of workload models - one-time workload model (i.e., simultaneous generation at the initial state) and continuous workload model (i.e., Poisson message arrival) - are considered. Our analysis reveals the effect of system parameters - the number of mobile agents on the field, the number of message loadings at a geographic location, the message generation rate and the number of message replicas - on the average and the distribution of message delivery delays. We also discuss the feasibility of a specific application of geographic DTN routing - communication among evacuation sites in disaster area.

In this paper, we derive the average message delivery delay in a geographic DTN routing with multiple mobile agents, where mobility patterns are given by random walk on a graph and message routing algorithm is the Random algorithm. A geographic DTN routing aims at realization of message delivery among multiple (generally, geographically-dispersed) geographic locations on a field without necessity of specific communication infrastructure by utilizing mobility of mobile agents. We model the behaviors of mobile agents as multiple random walks on a graph. Our analysis reveals the effect of system parameters - the number M of mobile agents on the field and the number K of message loads at a geographic location - on the average message delivery delay.

In recent years many public clouds have been available, and not only Infrastructure as a Service (IaaS) but others including Software as a Service (SaaS) as well have been provided by them. We focus on Amazon EC2 and RDS as examples of IaaS and SaaS in a public cloud, respectively, and experimentally compare them in terms of performance and costs in this paper. The results would be helpful for the users in considering which platform they should use for their applications.

A wireless LAN roaming system enables a user to access to a wireless network at a visiting institution with his/her account issued by his/her home institution. In the roaming system, an appropriate access privilege control is required. However, the conventional privilege authorization system gets unavailable because of some communication failures between institutions. In this work, we develop a fault-tolerant authorization system.

The appearance of various quality-sensitive applications has greatly changed the requirements for network management. To manage the quality of these applications, monitoring of individual traffic flows, as well as aggregated traffic statistics, has become more important. Since per-flow monitoring involves a high processing cost, especially in emerging high-speed links, packet sampling techniques have been attracting considerable attention. However, existing sampling techniques, such as NetFlow and sFlow, have mainly targeted traffic volume monitoring and there has been little discussion on the monitoring of quality indexes including packet loss rate. In this paper, we propose a method to estimate the TCP packet loss rate from sampled packets. The proposed method detects packet loss events by monitoring duplicate ACK events induced by a TCP receiver indicating the loss events. Since only a portion of packet loss events can be detected from the sampled packets, the correct packet loss rate is estimated by means of statistical approximation. Simulation results show that the proposed method accurately estimates the TCP packet loss rate from 10% of sampled packets.