Hideya Ochiai

A DTN-Based Sensor Data Gathering for Agricultural Applications

This paper presents our field experience in data collection from remote sensors. By letting tractors, farmers, and sensors have short-range radio communication devices with delay-disruption tolerant networking (DTN), we can collect data from those sensors to our central database. Although, several implementations have been made with cellular phones or mesh networks in the past, DTN-based systems for such applications are still under explored. The main objective of this paper is to present our practical implementation and experiences in DTN-based data collection from remote sensors. The software, which we have developed for this research, has about 50 kbyte footprint, which is much smaller than any other DTN implementation. We carried out an experiment with 39 DTN nodes at the University of Tokyo assuming an agricultural scenario. They achieved 99.8% success rate for data gathering with moderate latency, showing sufficient usefulness in data granularity.

Mobility entropy and message routing in community-structured delay tolerant networks

Many message routing schemes have been proposed in the context of delay tolerant networks (DTN) and intermittently connected mobile networks (ICMN). Those routing schemes are tested on specific environments that involve particular mobility complexity whether they are random-based or sociologically organized. We, in this paper, propose community structured environment (CSE) and mobility entropy to discuss the effect of node mobility complexity on message routing performance. We also propose potential-based entropy adaptive routing (PEAR) that adaptively carries messages over the change of mobility entropy. According to our simulation, PEAR has achieved high delivery rate on wide range of mobility entropy, while link-state routing has worked well only at small entropy scenarios and controlled replication-based routing only at large entropy environments.

FIAP: Facility information access protocol for data-centric building automation systems

Intelligent buildings are getting data-centric - they archive the historical records of motion detectors, power usages, HVAC statuses, weather, and any other information in order to improve their control strategies. The engineering cost of installation and maintenance of such systems should be minimized as the system owner has to operate them for several decades: i.e., the lifetime of the building. However, there are several design pitfalls that multiply such engineering costs, which make the operation heavy burden. This paper identifies those pitfalls and presents technical challenges that enable lightweight installation and maintenance. We, then, design facility information access protocol (FIAP) for data-centric building automation systems. We carried out FIAP-based system integration into a building of the University of Tokyo, and demonstrate that FIAP enables incremental installation for wide varieties of applications with small engineering costs.

Live E! Project: Establishment of Infrastructure Sharing Environmental Information

The Live E! project is an open research consortium among industry and academia to explore the platform to share the digital information related with the Earth and our living environment. We have getting a lot of low cost sensor nodes with Internet connectivity. The deployment of broadband and ubiquitous networks will enable autonomous and global digital information sharing over the globe. In this paper, we describe the technical and operational overview of Live E! project, while discussing the objective, such as education, disaster protection/reduction/recovery or business cases, and goal of this project activity

Building Hierarchical Switch Network Using OpenFlow

No single layers and technologies can construct a large scale network, thus networks are organized as a physical structure of variety of layers. For example in data center, networks are typically consists of access and aggregation L2 switches and core L2/L3 switches networks. However, new trends for server virtualization can change traffic dynamics, i.e. virtualized host can be located at any locations in the network and connections among servers are set up independently from physical network structure. Therefore static network structure cannot stay optimal and thus, we propose a network architecture whose structure is logically defined and flexibly reconfigurable. The network uniformly consists of OpenFlow switches and constructs a logical routing hierarchy defined by the software at OpenFlow controller. We evaluated our scheme to construct an efficient data center core network. We discussed types of logical structures and path selection algorithms with the constraint of flow table usage, which is one of the most critical constraints for scalable OpenFlow networks. Our simulation evaluation indicates that (1) full-mesh, hypercube and ring-based networks are the realistic logical structure design and that (2) hyperbolic path selection scheme has doubled the path capacity.

A field experience on DTN-based sensor data gathering in agricultural scenarios

This paper describes our field experience on data collection from remote sensors. By letting tractors, farmers and sensors have short-range radio communication device with delay-disruption tolerant networking (DTN), we can collect data from those remote sensors to our central database. Although, several implementations have been made by using PHS devices or mesh network in the past, DTN-based systems for such applications are still under explored. The main contribution of this paper is to present our practical implementation and experiences on DTN-based data collection from remote sensors. The software, which we have developed for this research, is very small - only about 3000 lines in C, which is much smaller than any other DTN implementations. We carried out an experiment with 10 DTN nodes in the University of Tokyo. They achieved 100% collection with moderate delivery latency showing sufficient usefulness in data granularity.

Facility Information Management on HBase: Large-Scale Storage for Time-Series Data

A very large number of sensors on facilities such as HVAC, light control systems and electric power meters, periodically submit their status information to Cloud platforms these days. As the amount of data can easily get petabyte scale, we must consider the use of distributed application layer storage for managing such facility information, which is often formatted on time-series data. This paper describes FIAP Storage Peta, petabyte scale storage for facility information access protocol (FIAP), proposing the architecture and the scheme of such data management on HBase. In this work, we have identified three requirements to the design of HBase row keys for implementing this storage using HBase. Though, we have not finished petabyte scale experiments, our preliminary evaluation results have shown good performance for managing large scale facility information. It has achieved scalable data retrieval on the data of 10 million sensors with properly balancing loads on distributed data storages.

Measurement Analysis of the Live E! Sensor Network: Spatial-Temporal Correlations and Data Aggregation

The (Japanese) Live E! project consists of a large sensor network of spatially distributed weather stations measuring different environmental quantities such as temperature, humidity, pressure, etc. Our goal is to conduct the first analysis on this huge data set of the structures of correlations both in time and space observed on data, for given quantity among stations, and for a given station, among sensors. Finally, we investigate a data aggregation algorithm, based on polynomial regression, and we show how it can reduce significatively the overall data traffic.

A Case Study of UTMesh: Design and Impact of Real World Experiments with Wi-Fi and Bluetooth Devices

This paper presents UTMesh -- a test bed for wireless mesh networking and delay (or disruption) tolerant networking developed in the University of Tokyo. We have customized 51 embedded Linux computers for experiment-oriented use cases, and let them powered by rechargeable batteries in order to flexibly deploy everywhere depending on the experiment scenarios. This paper summarizes the requirements for our test bed design, identifies an operation model of real world experiments, and presents a sample experiment and its results. UTMesh allows many types of real world experiments in a lightweight manner with Linux consoles, software importability and physical portability. This paper describes a link-level measurement study as a sample experiment. It has measured the differences of Wi-Fi multicast/unicast links and Bluetooth links. One of the lessons we must learn from the result is that Wi-Fi unicast sockets sometimes work poorly even if the node can communicate with multicast sockets especially when the network scale becomes large.

Hop-by-hop reliable, parallel message propagation for intermittently-connected mesh networks

Wireless mesh networks suffer from intermittent connectivity, and thus hop-by-hop reliability and parallel message propagation, which DTN researches have explored, can be applied to allow scalable message propagation over such challenged network environments. We implemented those communication schemes onto UTMesh - 50-node scale wireless mesh network testbed, and studied the delivery patterns. On the evaluation result with UTMesh, we have confirmed (1) that hop-by-hop reliability scheme achieves scalable message propagation (e.g., 23 hops), and (2) that message propagation speed increases as the redundancy-level increases. We have also observed that the smallest hop count path does not always achieve the fastest message delivery. This was probably because longer distant links were unstable and message paths over short distant links provided faster propagation.