Eiichiro Kodama

Assuring interoperability between heterogeneous multi-agent systems with a gateway agent

In this research, we address the problem of a lack of communications between heterogeneous multi-agent systems. For example, the KQML-based agent cannot talk to the FIPA-based agent and vice versa. We have identified three main reasons why different multi-agent systems are not interoperable, which we call architectural elements: (1) inconsistent mental state structures; (2) different syntax and semantics of the agent communication languages; and (3) incompatible message transport mechanisms. To solve this problem, we propose a gateway agent as a translator of the agent communications that assures the integration of heterogeneous multi-agent systems. We have developed a prototype of the gateway agent that can translate messages between the KQML agent and FIPA agent. Based on the results, we are studying a higher level of interoperability based on conversation protocols between agents.

Assuring interoperability in heterogeneous, autonomous and decentralized multi-agent systems

Multi-agent systems are autonomous decentralized platform software on which application programs such as business transactions, network provisioning and manufacturing control are deployed. In order for the applications to be truly autonomous and decentralized, however, heterogeneous multi-agent systems themselves must communicate and interoperate with each other. To solve the problem, we take a two-step approach. First, the message-level interoperability is achieved by a gateway agent that interconnects heterogeneous multi-agent systems. Next, a higher-level interoperation of conversations, which consist of bi- (or multi-) directional streams of messages, is governed by interaction protocols. We demonstrate the concept and technique of dynamic negotiation of interaction protocols and show how the method improves the assurance of real-world applications in autonomous decentralized systems.

Reducing data leakage possibility resulted from eavesdropping in wireless sensor network

Wireless sensor networks are expected to be used in a variety of fields ranging from military applications to personal health monitoring. Due to the constraints in computation, memory, and power resources, and because of the nature of the shared transmission medium which makes it relatively easy for attackers to eavesdrop, avoiding information leakage is a challenge. This paper proposes a routing-based approach against the eavesdropping to enhance sensor network security. Simulation experiment results have confirmed the advantage of the new routing algorithm over existing ones.

A Scalable Sequential Pattern Mining Algorithm

Sequential pattern mining is a technique used to discoverfrequentsubsequencesaspatternsina sequencedatabase. Many excellent sequential pattern mining approaches such as GSP, SPADE, and PrefixSpan have been proposed. However, the existing approaches still encounter problems when the set of all different items in a sequence database is large. Scalability with respect to not only the varying size of a sequence database, but also the varying size of the set of all different items is crucial for many applications. In this paper we address the subject of mining frequent sequential patterns in the large sequence database with numerous kinds of items. An effective algorithm for the purpose, called SSPM (ScalableSequentialPattern Mining), is proposed. SSPM is characterized by its fast convergence, meaning that the search space will shrink quickly as mining operation proceeds. SSPM does not limit the number of different items in a sequence database, and thus it has better scalability, and can be applicable to a large scale of sequence databases. The experimental and analytical results demonstrated that SSPM is effective and is faster than conventional algorithms in the above stated cases.

A Distributed Approach to Constructing k-Hop Connected Dominating Set in Ad Hoc Networks

Since ad hoc networks do not have fixed or predefined infrastructures, nodes need to frequently flood control messages to discover and maintain routes, which causes performance problems in terms of unnecessary traffic and energy consumption, contention, and collision. A general solution is to construct a virtual backbone as the basis of routing and broadcasting, and the Connected Dominating Set (CDS) has been widely used. This paper gives a distributed approach to constructing k-Hop CDS with three unique characteristics: (1) the limitation on the range of k has been removed, (2) a token-based conflict avoidance mechanism has been introduced which can make the construction process faster and more effective, (3) a tree-type CDS can be constructed in bottom-up processing. Simulation experiments have been conducted to demonstrate the effectiveness of the proposed approach.An Image Signal Processor (ISP) converts raw imaging sensor data into a format appropriate for further processing and human inspection. This work explores FPGA-based ISP designs considering specialized and programmable implementations and proposes an ISP using a programmable generic processing unit with comparable performance versus the dedicated implementations.Hadoop as a popular open-source implementation of MapReduce is widely used for large scale data-intensive applications like data mining, web indexing and scientific computing. The current Hadoop implementation assumes that nodes in a cluster are homogeneous in nature, and Hadoop distributed file system(HDFS) distributes data to multiple nodes based on disk space availability. Such data placement strategy is very efficient for homogeneous environments, where nodes are identical in terms of both computing power and disk capacity. Unfortunately, in practice, the homogeneity assumptions do not always hold. Hadoops scheduler will lead to severe performance degradation and energy dissipation in heterogeneous environments by using default data placement strategy of HDFS. In this paper, we propose a novel snakelike data placement mechanism (SLDP) for large-scale heterogeneous Hadoop cluster. SLDP adopts a heterogeneity aware algorithm to divide various nodes into several virtual storage tiers(VST) firstly, and then places data blocks across nodes in each VST circuitously according to the hotness of data. Furthermore, SLDP uses a hotness proportional replication to reduce disk space consumption and also has an effective power control function. Experimental results on two real data-intensive applications show that SLDP is energy-efficient, space-saving and able to improve MapReduce performance in a heterogeneous Hadoop cluster significantly.

Mining context-related sequential patterns for recommendation systems

A typical recommendation system answers such questions as what are the interesting items for the current user. Most traditional recommendation systems have not taken the situational information into account when making recommendations, which seriously limits their effectiveness in the ubiquitous computing application environment, where a users request is generally related to, and a systems response should be dependent on, a specified context (e.g., a specific place, time slot, noise level, or temperature range). In this paper we propose a context-aware recommendation approach to enhance the performance of recommenders. This approach is characterized by a novel sequential pattern mining algorithm that can efficiently mine and group patterns by context.

Construction and Maintenance of K-Hop CDS in Mobile Ad Hoc Networks

Since mobile ad hoc networks do not have fixedor predefined infrastructures, nodes need to frequently floodcontrol messages to discovery and maintain routes, whichcauses performance problems in traffic and energy consumption, contention, and collision. A general solution is to constructa virtual backbone as the basis of routing and broadcasting, and the Connected Dominating Set (CDS) has been widely used. This paper presents a distributed approach to constructingand maintaining a k-hop CDS. A unique characteristic thatdifferentiates the proposed approach from existing ones is that, any value can be specified for k, and a larger k contributesto a smaller CDS. If k is large enough, the resulting CDSwill contain only a single node. Simulation experiment resultsdemonstrate the effectiveness of the proposed approach.

An Onomatopoeia-Based Web Music Video Searching System and Its Performance Evaluation

In recent years, as use of the Internet became widespread, numerous music videos became available on Web. In Japan, many of these music videos are the CGM (Consumer Generated Media) that are created using a singing synthesis software called Hatsune Miku, and published on YouTube and other similar Web sites. Existing Web sites, however, support only the search methods based on music video title and artist name, which could not be effectively used to search for the unknown music videos such as the CGM ones. This paper presents a system model for effectively searching for the unknown music videos, which is characterized by the use of the onomatopoeia. The system model consists of a music video collecting engine for collecting pairs of music video URL and its tags, an onomatopoeia assigning engine for assigning onomatopoeias to music videos, and an onomatopoeia retriever for presenting users the music video URLs satisfying their onomatopoeia requirements. We have implemented a prototype system of the proposed system model and conducted experiments to study its performance. It has been found that with the proposed system model, a precision ratio of 66.82%, a recall ratio of 56.36%, and an F-measure of 61.14% could be achieved.

Sleep thrashing and its prevention in ad hoc networks

In the disaster-hit area, communication supporting systems using the existing mobile ad hoc network technique could not function as expected and communication would be interrupted from time to time, since they have not taken into consideration the issue of sleep, which is a very frequently used means for users to save their battery consumption. Especially, a phenomenon called sleep thrashing may occur, which would collapse a mobile ad hoc network. This paper explains the concept of sleep thrashing, and proposes a system model and a CDS-based solution for reducing its negative effect on communication. Simulation experiments have been conducted to demonstrate the effectiveness of the proposed approach.

Supporting Communication in Time of Disaster

It was found that at the disaster-hit area, communication supporting systems using the existing mobile ad hoc network technique could not function as expected, and a communication process between two end users would be interrupted from time to time. This is because existing techniques have not taken into consideration the sleep operation of mobile users, which is a very common and frequently-used means for them to save their battery consumption. This paper proposes a routing-based approach to solving this problem. For the proposed approach, power management application and routing protocol are incorporated so that mobile users can help each other by technical means to maintain the network accessible. This approach has been demonstrated to be effective by simulation experiments.