Search

In this paper, we present Genetic Algorithm based optimized feature selections for intrusion detection systems. We used one-point crossover for the Genetic Algorithm parameters instead of two-point crossover used by the previous research as it one-point crossover is faster. For evaluations, we used the NSL-KDD Cup 99 data set and we modified the data set by looking into to the recent attacks, hence making the data set more relevant to the current situations. Several classifiers were used on these data sets and we found that Random Forest gave the best results in terms of the classification rate and the training time. The results also showed that our parameters performed better in these two metrics and the classifications using our optimized features on the modified data sets gave mixed results compared to ones with the original features.

Applications requiring strict access control, such as medical record query, often require auditing of the query. The current typical design relies on server side logging. However, logging on server-side do not provide strict means of auditing, since the server can be tampered with attackers, and also anybody who has permission to write can modify the log. We propose a scheme using blockchain technology, as a request-response channel for a client-server system, to record both client request and server reply in an audi-table manner. We have implemented a proof-of-concept system on top of a publicly available blockchain testbed. By using a blockchain as a client-server request-response channel, the request-response sequence can be verified by anybody who has access to the blockchain, providing a way to implement audit log for strictly controlled resources.

Occupancy determination of a large room, such as university campus room, using pyroelectric infrared (PIR) sensors is difficult because PIR sensors are inherently insensitive to humans that are motionless and it is difficult to achieve 100% coverage even with multiple sensors. To overcome these problems, the authors propose using particle filtering of networked multiple PIR sensor data to improve the accuracy of the occupancy state determination. The particle filtering can eliminate incorrect occupancy state transition without incurring large delays. Our method is robust to the incompleteness of the coverage and accuracy. The authors performed an experiment in a campus room (15.3m×6.4m) with 4 PIR sensors to prove the validity of the proposal. The experiment shows that the proposal can improve the occupancy determination accuracy from 29.6% to 98.3%. The system has been continuously used in the university power saving project.

Physical objects often form a group such as objects in a shipping container. RFID enables us to identify each object and even the container itself. However, current RFID does not provide information on IDs missing from a group. This paper proposes a method to determine the unique IDs of objects missing from a group without any external database or verifier. The proposed method logically splits a group into mutually overlapped sub-groups and writes group-related information, which is generated from the unique IDs of objects in the subgroup, to RF tags' memory. When we check the integrity of a group of objects, unique IDs and group-related information of RF tags are extracted from RF tags' memory. With an iterative decoding over group-related information with the unique IDs of identified objects, missing IDs are determined. A numerical simulation reveals that the proposed method can identify 96-bit unique IDs of up to 64 objects missing from a group composed of 100 objects by writing 840-bit group-related information to each RF tag. We also examined the performance with an experiment and confirmed that we can successfully determine 16-bit IDs of up to 12 missing RF tags from a group of 20 RF tags by writing 280-bit group-related information to each RF tag. The experiment results agree well with the numerical simulation.

In this paper, iDANS, which is a location based information dissemination platform for smartphones composing a vehicular ad-hoc network, is proposed and evaluated. The primary objective of iDANS is to disseminate and share information swiftly between vehicles within a designated area. In order to meet the demands, iDANS disseminates the information which is most relevant to the current location on a priority basis. It also makes use of devices that cannot acquire the current location accurately as relay nodes. The technical feasibility of iDANS was verified on iPhones, and simulations were done in order to verify the effectiveness of the platform. As a result of the simulations, it was verified that iDANS was up to 5.5 times more effective compared to random dissemination, in terms of disseminating more information among vehicles within a designated area.

RFID is an essential technology to uniquely identify physical objects. In many practical business processes using RFID, we check not only individual objects but also check if there is no missing objects and no extra objects in a group of objects. While individual objects identifications are done by using RFID, group verifications are usually done by looking up a shipping list or corresponding EDI data which require a network connection. In this paper, we propose “group coding of RF tags” by which we can verify the integrity of a group of objects just by writing additional data in RF tags' memory. The additional data is computed from unique IDs of objects that belong to the group. We propose fundamental and general group codings. With the fundamental group coding, we can check if the integrity of a group of objects is preserved or not. It is a parity check of hashes of unique IDs in a group. We also propose general group coding by extending the fundamental group coding. With the general group coding, we can estimate the number of missing objects if the integrity of a group is not verified. The strength of the missing number estimation can be controlled by the size of data written to RF tags. It can be considered as a Low Density Parity Check (LDPC) in physical objects. The theory of group coding is confirmed by a numerical simulation and an experiment. It is shown by the simulation and experiment that we can detect 10 missing RF tags out of 20 RF tags with 99.5% reliability by writing 96-bit data into each RF tags besides its 96-bit unique ID.

This paper describes an experimental study of the overlay topologies of real-world Bittorrent networks, focusing on the activity of the nodes of its P2P topology and especially their dynamic relationships. Peer Exchange Protocol (PEX) messages are analyzed to infer topologies and their properties, capturing the variations of their behavior. Our measurements, verified using the Kolmogorov-Smirnov goodness of fit test and the likelihood ratio test and confirmed via simulation, show that a power-law with exponential cutoff is a more plausible model than a pure power-law distribution. We also found that the average clustering coefficient is very low, supporting this observation. Bittorrent swarms are far more dynamic than has been recognized previously, potentially impacting attempts to optimize the performance of the system as well as the accuracy of simulations and analyses.

We have developed a system using a combination of a multi-hop sensor network and 3G networks to support patients daily medication in home care. Each medication of a patient is represented by a sequence of close-open-close events of an off-the-shelf pill box, which are detected by a small Hall sensor. The system enables home visiting nurses to remotely monitor their patients' medication and to give feedback on the medication status to patients. This bidirectional communication between medical staff and patients is essential to motivate patients' medication and to make their experience of being taken care for at homes more comfortable. A semi-automatic message generation system helps medical staff to send feedback without spending excessive time and effort. We have been examining this system for more than three months with three subjects and found that a simple open/close sensing of an off-the-shelf pill box can provide a valuable medication history. It has also shown that the feedback system improves the rate of failure medication from 5.3% to 0.0% in the best case.

This study proposes RFID data analytics methods using consumer purchase behavior data as well as sales data. Consumer purchase behavior data, how consumers select or not select items at stores, has not been available, but with the progress of the radio frequency identification (RFID) technology, it becomes feasible to use it in a commercial basis. The proposed data analytics methods are two: one is to infer items that are compared to each other while customers select items at stores, and the other is to infer good combination items that can be used to stimulate sales of each other. We evaluate our proposal using real consumer purchase behavior data and sales data captured in an RFID pilot experiment. We confirm that the methods successfully infer the items that are useful to the retailer.

UHF band passive RFID is an essential identification technology in pervasive computing world. One of the industrial demands to UHF band passive RFID is fast reading of many RF tags. This is particularly relevant for item-level-tagging of CDs, DVDs, books and consumer electronics. Many high-speed MAC protocols have been proposed thus far. Quantitative evaluation of reading speed, however, is not easy because we need to lay out many RF tags and measure the reading durations usually through interrogator APIs, which differ from interrogator to interrogator. In this paper, the authors propose two efficient multiple RF tag emulation algorithms which leverage the probabilistic nature of MAC protocol. We have implemented the algorithms onto a programmable battery assisted passive tag (BAP), equipped with an 8bit micro processor and evaluated the reading speed of commercial interrogators. It was revealed that the BAP can efficiently emulate the behavior of more than 1000 Gen2 RF tags for the reading speed evaluation with a trivial penalty on accuracy.

Consider a situation where one desires to retrieve sensor data from an arbitrary environment via a moving interrogator. Maintenance of street lights and electric apparatus in conduits for example. The fundamental problems with this system have been 1)the trade-off relation between power consumption by sensor and the data retrieval rate for the interrogator, and 2)the data retrieval agility for the moving interrogator. A higher duty ratio incurs more power consumption in sensors whereas a lower duty ratio leads to a less reliable retrieval rate. Even though there have been such efforts as" adaptive sleep period control "which was derived from the packet arrival rate, they didn't systemically solve the problem, particularly for moving-interrogator environment. This paper presents a solution of separating the timing of sensor-sampling and the timing of sensor-data retrieval. The asynchronous data retrieval mechanism is facilitated by a dual interfaced UHF passive RF tag known as a "recorder tag" which the authors have been developing. This paper examines the power consumption composition of a prototype recorder tag. The power consumption efficiency of the proposed data-retrieval method is discussed in comparison with that of the existing method in terms of "Sensor Sampling Interval vs. Battery Life Expectancy." The authors also evaluated the viability of retrieving sensor data through a moving-interrogator. We specifically inspected the relationship of "Interrogator's Moving Speed vs Successful Data Retrieval Rate." The tests showed that with a commercial 1.2V rechargeable battery, the recorder tag lasts for 21 days forluminance and temperature sensing using a 30 minute samplinginterval, extending the traditional battery lifespan ten times longer, yet still preserving 100% time availability for sensor data retrieval. The power consumption is expected to be eight times as efficient when using optimized circuitry. Experiments in an emulated fading environment reveal that the data retrieval rate is observed to be the highest when the maximum relative speed between the moving-interrogator and the recorder tag is less than 3km/h, with an optimized distance of separation being 10cm for a 27dBm EIRP interrogator. The moving speed, 3km/h, can be improved by optimizing the anti-collision procedure and by enhancing the sensitivity of the recorder tag.

We have implemented Multi-Session based Network Security Event Detector: ROOK to detect botnet activity and P2P file sharing traffic and our results show that our method is less false positives than existing network security event detectors (e.g. IDS). We proposed a network security event detection method by analyzing correlation among multiple sessions. Our method can recognize hosts behaviors by rules that describe multi-session correlations: a rule includes the order of starting sessions and information exchange between sessions. By this method, ROOK detected DNS and IRC activities of bots in the experiment.

In this paper, we present a new computer architecture called All-IP computer architecture, which realize a flexible computer re-configuration environment via IP networks. Our new IP-based protocols connect devices to the host computer that were previously limited to a physical attachment. To construct All-IP computer environment, we designed a protocol to manage and discover device resources on IP networks. We also evaluated a protocol called USB/IP, which is expected to be one of the device interconnection protocols in the All-IP computer architecture.

This paper discusses the problems of conducting a real-time workshop, where the instructors and the participants are located in different places, and the participants should perform hands-on lab exercises using computers that are located in another place. To address this problem, we implemented a remote hands-on exercise environment using virtual machines and a low-bandwidth remote console access. In this environment, the instructors can monitor each participant's exercise progress thus they can give specific feedbacks to the participants. We evaluated this environment using a real-time online workshop. We found that it was easy to setup, and participants had a good user-experience using this environment

This paper proposes an adaptive Overlay Multicast protocol for real-time group communication in a heterogeneous environment. Overlay multicast technology relies on end nodes by delegating the multicast functionality from routers in IP Multicast. Overlay multicast research has two major issues to consider: heterogeneous resource environment and instability of end nodes. In our protocol, data is divided into multiple layers using abstract layered data structure. The number of layers is used for the main metric to construct the multicast tree to satisfy the demand for end nodes resource environment. Furthermore, multi-path layer distribution method for fast recovery from multicast tree partition and congestion avoidance method are proposed to deal with the unstableness at end node.

Peer-to-peer (P2P) is a system of overlay networks such that participants can potentially take symmetrical roles. This translates itself into a design based on the philosophy of local production, local consumption (LPLC), originally an agricultural concept to promote sustainable local economy. This philosophy helps enhancing survivability of a society by providing a dependable economic infrastructure and promoting the power of individuals. This paper attempts to put existing works of P2P designs into the perspective of the five-layer architecture model to realize LPLC, and proposes future research directions toward integration of P2P studies for actualization of a dependable and sustainable social infrastructure

This paper proposes an alternative model to access streaming video contents available on the Internet. Currently many streaming videos are delivered in one way, in spite of the bi-directional network connection. To have more interactivity in use of them, we have implemented a system for viewers to control the original unedited video sources with active viewing interfaces. The system allows video source owners to share video sources without editing, and viewers to reorganize them while receiving from the streaming server. With the implemented system, viewers can utilize the video sources in more effective way for educational purposes

E-learning has made possible to conduct trainings online. However, region-wide trainings that require hands-on practice still need lecturers and participants at the same location. This requires huge resources on travel budget, time, equipments and staff. This paper proposes a region-wide distributed model of realtime hands-on workshop using distance learning and remote virtualization environment. The proposed model was implemented for SOI Asia operator training during 22-26 August 2006 inviting 42 participants from 19 institutions in 10 Asian countries. The workshop proved to have significant reduces in travel budget, lab preparation time and equipments compared to ordinary workshop model held in the previous year. Feedbacks show 95% of the participants evaluated this workshop is effective for their future operations. Accordingly, this workshop model is practical and resource-optimized model for region-wide human resource development. This paper can be a useful information source for future development and improvement of region-wide remote workshops

Post-catastrophic recovery is gaining more importance to human societies. This paper proposes use of a distributed autonomous economic medium to support recovery of communities after catastrophic events. It proposes a model in which everyone in the world can help each other as peers, instead of creating one's dependencies on others. The model applies peer-to-peer technologies on the Internet to achieve an environment for mutual help in the planetary scale

SOI Asia project aims at Asia-wide distance education development with its design emphasizing on new model of network technologies that provides wide-coverage high-quality network to support high-quality distance education instead of relying on the available Internet connectivity with inadequate quality. A 13-Mbps unidirectional broadcast satellite covers Asia region enabling lecture sharing among student sites. IPv6 multicast overlay network is established by UDLR and IPv6 tunnel technologies to create a virtual network connecting lecturer-site and student-sites on top of network heterogeneity. Bandwidth optimization and scalability are achieved by IPv6 multicast. QoS using policy routing and ALTQ utilizes and manages bandwidth to ensure high quality lectures. Experiment results show that SOI Asia environment delivers better quality streams with stable rate, high correlation, low data loss and jitter compared to the available Internet connectivity. These design aspects, technology model and experimental results can be the considerations of future region-wide distance education development

A peer-to-peer complementary currency can be a powerful tool for promoting exchanges that make use of underutilized computing resources in a trusted way. i-WAT is proposed such currency based on the WAT System, a polycentric complementary currency using WAT tickets as its media of exchange: participants spontaneously issue and circulate the tickets as needed, whose values are backed up by chains of trust. i-WAT implements the tickets electronically by exchanging messages signed in OpenPGP [J. Callas et al., (1998)]. This paper proposes an extension to the design of i-WAT to facilitate trades. In particular, we propose additional "multiplication" tickets whose values increase over time. By deferring redemption of such tickets, the participants can receive the premium realized by increased debts of the issuers. A game-theoretical analysis shows that this feature is incentive-compatible: the issuers have no reason to strategically default despite the increased debts. A reference implementation of i-WAT has been developed in the form of a plug-in for an XMPP instant messaging client. We have been putting the currency system with the proposed feature into practical use.

In the near future, mobile communication will become majority on the Internet according to quick progress of wireless technology and mobility protocols. Mobile IPv6 has been standardized at IETF and almost ready to deploy on the Internet. However, handover technology of Mobile IPv6 is still under development. Since Mobile IPv6 does not specify optimized handover mechanism, it takes certain period to complete handover. In this paper, we investigate a mechanism to minimize handover latency without any modification to Mobile IPv6. We take position that a mobile node will utilize multiple interfaces to achieve smooth handover. With multiple interfaces, the Internet access can be continued even while Mobile IPv6's movement is proceeded. SCTP is a transport protocol capable of handling multiple addresses for each session. A mobile node uses a home address as a primary address and another available address as a secondary address. The SCTP fails over to the secondary address while the primary address (i.e. home address) is inactive during handover. The advantage of this proposal is that it required modification to neither Mobile IPv6 and SCTP.

The WIDE project, www.wide.ad.jp, is a research consortium among industry and academia. WIDE project consists of more than four hundred active researchers from more than hundred private companies and from more than forty universities. The WIDE project operates the nationwide R&D testbed, which is connected with many other R&D testbeds, and the NSPIXPs, that are the largest IXes in Japan. The main focus of WIDE project has been IPv6 technology and has established IPv6 testbed since 1998. Basically, all of R&D activities in WIDE project are based on the IPv6. This paper describes the overview of WIDE project's IPv6 R&D activities, such as KAME/USAGI/TAHI/Nautikus6 projects.

Network Mobility (NEMO) Basic Support protocol is the standard to provide continuous network connectivity and movement transparency to a group of nodes moving together, as in a vehicle. However, the protocol suffers from suboptimal routing and packet overhead caused by a bi-directional tunnel between the Mobile Router (MR) connecting the mobile network to the Internet and its Home Agent (HA). When nested mobile networks are formed, these issues are critical for real-time appli- cations. This study proposes a mechanism which provides suffi- cient optimal path applicable to the various configuration known in nested mobility. The proposed Optimized NEMO (ONEMO) protocol decreases packet overhead, uses minimum signaling overhead, and modification is limited to MRs. Evaluation through comparison among other solutions showed effectiveness of the scheme.

This paper looks at the message implosion problem that may happen on large-scale reliable multicast communications. We propose that receivers track the senders of recent feedback messages and store them in a list to determine the back-off interval to send a feedback message. We call our proposal tracked sender list (TSL). The issues regarding this proposal are how many receivers to be stored in a receiver's list, back-off timer intervals, p.d.f selection, and how a receiver stores itself in its list. We analyzed and simulated our proposal using various parameters values in both networks with equal delay and with heterogeneous delays to show the message suppression performance. The results demonstrate that TSL gives a low number of messages over a wide range of receivers even with a short back-off interval and it is not susceptible to packet losses. Our results also show that TSL performs better than other p.d.fs that is used in message suppression.

This paper proposes a novel technique to enable flow-level multi-path forwarding on IPv6 networks. Generally on such networks, any packet from a node is forwarded toward a destination along the same path, even if some of the packets belong to a different flow with different characteristics. In accordance with packet characteristics and to fulfil demands such as traffic engineering and QoS routing, flow-level multi-path forwarding allows explicit selection of the paths on which to send packets. Multi protocol label switching (MPLS) technology has been used to perform this function so far. However, to apply MPLS on an existing IPv6 network, all non-MPLS-enabled routers in the network must be replaced with MPLS-enabled routers. Therefore, the migration cost is generally expensive. Our previous work has proposed a technique called flow-mapping explicit host routing (FMEHR,), that renders similar results to MPLS using unused address spaces such as IP private addresses. This paper describes the FMEHR concept and its realization on an IPv6 network and discusses its characteristics.

A P2P currency can be a powerful tool for promoting exchanges in a trusted way that make use of under-utilized resources both in computer networks and in real life. There are three classes of resource that can be exchanged in a P2P system: atoms (ex. physical goods by way of auctions), bits (ex. data files) and presences (ex. time slots for computing resources such as CPU, storage or bandwidth). If these are equally treated as commodities, however, the economy of the system is likely to collapse, because data files can be reproduced at a negligibly small cost whereas time slots for computing resources cannot even be stockpiled for future use. This paper clarifies this point by simulating a small world of traders, and proposes a novel way for applying the "reduction over time" feature of i-WAT, a P2P currency. In the proposed new economic order (NEO), bits are freely shared among participants, whereas their producers are supported by peers, being given freedom to issue exchange tickets whose values are reduced over time.

In this paper, virtual mobility control domain (VMCD) is proposed to enhance mobility protocols such as mobile IP and network mobility (NEMO). These mobility protocols rely on a single agent called the home agent. This leads to protocol weakness in terms of system robustness and redundant routes. In VMCD, multiple anchor points serving the same home network are placed as home agents at different networks on the Internet backbone. These anchor points advertise the route of the same home network to the backbone from different networks. It is similar to anycast routing approach of root DNS. By genuine routing, packets meant for mobile nodes are always drawn into the closest anchor point and are delivered to the mobile node by the anchor point on behalf of home agent. In addition, even if one of the anchor points fails, another anchor point immediately takes over for the failed anchor point. VMCD provides better routes for mobile nodes and protocol robustness with minimal involvement of end nodes. This paper discusses the VMCD solutions and shows the efficiency and effectiveness of VMCD.

Internet technology assumes all links are bidirectional, hence it does not take advantage of the broadcast nature of radio frequency communications. RFC3077 based on E. Duros et al. (2001) emulates broadcast connectivity in UDL (UniDirestional Link) and it provides generic solution for all network layer mechanisms. However, this method doesn t support some OSPF based on J. Moy (1998) network topology. We extended the definition of UDL. Our extention enables a node does not have to be reachable from another node. With our extention, we can use every connectivities over UDL as routes then we extended OSPF to handle unidirectional links. Our extension enables all OSPF routers to construct the routing table including UDL, and it is compatible with general OSPF implementation.

This research focuses on a contribution to high quality video and audio transport system using network infrastructure. Our objective is to enhance the realization of higher communication environment using easy high quality transport application. Basic structure of DVTS (Digital Video Transport System) is based on transporting DV (Digital Video) through the Internet. DVTS realizes cost effective solution using DV, to perform high quality audio and video communications. Our research goal of this year is to enhance usability of DVTS by implementing multiple audio transmission system, portability enhancement to DVTS, Multi point, real time editing capability, and implementation of MPEG2-TS. We established a new communication model using DVTS as a base of communication method.

RFID attracts attention as a useful device in location-aware application. SPEARS WG of WIDE Project had already developed a system (2003), which uses RFID devices as a sensor of human's location. Purpose for this system is to provide the service based on a user's location in a conference. In this system, the model in which location of user is gathered and provided for each user was adopted. This paper describes several areas of problems and improvements in our latest implementation. The following functions were implemented by the addition of a system called "Area-d": 1) provide a service tied with a location, and 2) provides users' communication rendezvous service. These functions are able to add to the existing system without altering the existing model. This paper describes a location-oriented data processing model and applications.

In this paper, we describe the Optimized Route Cache Management Protocol (ORC) as a network mobility protocol and evaluated it through experiments on a prototype implementation. ORC provides mobility transparency to a network regardless of network movements. Mobile network is a network that moves entirely. Typical examples for a mobile network are personal area networks (PAN) and networks inside vehicles. ORC is based on mobile IPv6 and is combined with the Internet routing mechanisms for network mobility management. ORC assigns a unique unchanging subnet prefix to a mobile network and notifies and maintains a mobile network's route called binding route (BR). A BR is an association between the prefix of the mobile network and a current available address known as care-of address in mobile IPv6. ORC allows interior gateway protocol (IGP) routers named ORC routers to cache a BR. ORC routers intercept packets destined to the mobile network and route them to the mobile network according to the BR. The BR must be updated as soon as the mobile network changes the care-of address by its topological position, employing constant BR exchange to ORC routers and temporary BR exchange to dynamically discovered ORC routers. Based on the experimental results, ORC is confirmed to provide network mobility and optimal communication in terms of route optimization. The evaluation shows that route optimization is required for network mobility to improve communication performance.