A Public Vehicle-based Urban Sensing System - ACM Digital Library

14 downloads 0 Views 1MB Size Report
Hideyuki Tokuda. Keio University. 5322 Endoh. Fujisawa, Kanagawa, JAPAN [email protected]. Permission to make digital or hard copies of part or all of this ...
UBICOMP/ISWC '15 ADJUNCT, SEPTEMBER 7–11, 2015, OSAKA, JAPAN

A Public Vehicle-based Urban Sensing System Yin Chen Keio University 5322 Endoh Fujisawa, Kanagawa, JAPAN [email protected]

Yutaro Kyono Keio University 5322 Endoh Fujisawa, Kanagawa, JAPAN [email protected]

Takuro Yonezawa Keio University 5322 Endoh Fujisawa, Kanagawa, JAPAN [email protected]

Jin Nakazawa Keio University 5322 Endoh Fujisawa, Kanagawa, JAPAN [email protected]

Kazunori Takashio Keio University 5322 Endoh Fujisawa, Kanagawa, JAPAN [email protected]

Hideyuki Tokuda Keio University 5322 Endoh Fujisawa, Kanagawa, JAPAN [email protected]

Abstract In this poster, we propose a public vehicle-based urban sensing system, where sensor nodes are equipped to public vehicles and collect data as the host vehicle roams around a city. Compared with the urban sensor systems that employ static sensor nodes or personal mobile devices, our proposed system achieves a wide sensing area through the mobility of host vehicles, reduces initial investment by employing existing infrastructures and facilitates system operation and maintenance which can be executed in the car barn(s). In this poster, the system architecture and implementation details are described. A preliminary experiment is conducted to validate the proposed system.

Author Keywords Urban Sensing; Sensor Network; Vehicular Network.

ACM Classification Keywords H.4.m [Information Systems Applications]: Miscellaneous. Permission to make digital or hard copies of part or all of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for third-party components of this work must be honored. For all other uses, contact the Owner/Author. Copyright is held by the author/owner(s). UbiComp/ISWC’15 Adjunct, September 7-11, 2015, Osaka, Japan. ACM 978-1-4503-3575-1/15/09. http://dx.doi.org/10.1145/2800835.2800886

117

Introduction An urban sensing system is a collection of sensor nodes deployed over a city to collect interested data, which will be processed to produce information. It is envisioned that these information will be useful in city trends prediction, city management and design, service performance improvement, etc. One approach to implementing such an

UBICOMP/ISWC '15 ADJUNCT, SEPTEMBER 7–11, 2015, OSAKA, JAPAN

urban sensing system is to firstly deploy sensor nodes in the interested areas and then gather the generated data via either a communication network or mobile data collector(s). While the pre-installed static sensors are able to constantly monitor an area of interest, implementing such a distributed urban sensing system over a large area requires a great amount of initial investment and the operation and maintenance will be also very timely and financially expensive. As a low-cost solution, some researchers proposed to implement urban sensing systems by exploiting mobile phones, a kind of ubiquitous device in the modern society that is integrated and/or can be connected wirelessly with sensors [1]. While the mobile phone-based approach allows human participation in the sensing process and thus makes it possible to collect subjective data, it is not practical to collect environment data of cities due to the limitation of sensors equipped/connected to mobile phones. Besides, the participators’ concern of privacy leakage is also a barrier in the application of such a system. This work proposes a public vehicle-based urban sensing system, where sensor nodes are equipped into public vehicles that roam over the city for their duties. Examples of public vehicles include, but not limit to, waste collection truck, public transport bus and police car. In such a system, the equipped sensors are able to fulfill a wide coverage as the host vehicles typically roam over the main areas of a city. Compared with the distributed static sensing system, the operation and maintenance will be greatly facilitated in the proposed system, since they can be conducted after the host vehicles returned to their base. Besides, since the public vehicles are typically owned by local governments, the major investors and potential users of urban sensing systems, using existing infrastructures and reducing initial investment and

118

operational cost should be of great interest to them. In our study, a prototype system was implemented into the waste collection trucks of Fujisawa city, Japan, bases on which we also conducted a preliminary experiment to verify the applicability of the system.

Figure 1: The System Architecture

System Overview We adopt a client-server architecture for the urban sensing system. As shown in Figure 1, the system consists of a data center server, several sensor nodes and the same amount of their host public vehicles (i.e., waste collection trucks in this study). Data Center Server The data center server is named Sensor-Over-XMPP (SOX), an open-source platform developed by Hide Tokuda Lab. of Keio University which can publish and subscribe sensor data based on the Extensible Messaging and Presence Protocol (XMPP) protocol [5]. The readers are referred to [3] for more information.

POSTERS

the data center server via the wireless router.

Figure 2: Sensor Node. Figure 3: Sensor Module.

Figure 4: Mini Computer: Raspberry Pi.

Figure 5: 4G-LTE Wireless Router.

Sensor Node As shown in Figure 2, the sensor node consists of a sensor module, a mini computer and a 4G-LTE wireless router [4]. The sensor module (Figure 3) is integrated with some typical environmental sensors including a temperature sensor, a humidity sensor, a ultraviolet light sensor, a dust density sensor as well as a GPS receiver that can measure the current location of the node. The mini computer is a Raspberry Pi Model B + (Figure 4) running RASPBIAN (i.e., a linux distribution) and Java Running Environment 1.8. A usb wifi dongle is equipped to the mini computer and used to access the Internet through the 4G-LTE wireless router. The 4G-LTE wireless router uses the LTE cellular network of NTT Docomo, one of the major mobile network operators in Japan and whose LTE service area covers the whole Fujisawa City. The sensor node works as follows. Firstly, the sensor module periodically reports its generated sensor data (including GPS coordinates) to the mini computer through a serial port. Secondly, a self-developed Java program is running on the mini computer. It constantly reads data from the serial port, transforms the data into the required format and finally sends them in real-time to

119

Waste Collection Truck In this study, we equipped our developed sensor node to the waste collection trucks of Fujisawa city of Japan. These trucks work from 8:00 am to 15:00 pm in every weekday. They move along pre-determined routes and collect waste from the waste containers along the way. In Japan, each house/mansion has at least one waste container, to which the residents of the building will throw their family waste following the schedule made by the corresponding local government. In our case, the waste collection trucks of Environment Business Center of Fujisawa are divided into 7 groups. The first 6 groups conduct routine collection, each of which is assigned to a specific part of the city. The last group is called ”Illegal Disposal Prevention Truck”, the members of which patrol to the places where illegal disposals occur frequently. These places are typically far way from residential areas since the suspects of illegal disposal tend to avoid from being witnessed. To sum up, the waste collection trucks travel over most area of the Fujisawa city. Protocol Stack and Data Format As shown in Figure 1, the adopted communication protocol stack consists of LTE link layer, TCP/IP layer and XMPP layer. XMPP is a message-oriented communications protocol based on XML and it enables the near-real-time exchange of structured extensible data and thus has been widely applied in sensor networks. The data sent to the SOX server are text-based and each data message consists of three elements: raw value, typed value and timestamp. The raw value represents the reading directly obtained from the sensor module. The typed value represents a human readable representation of the data. The timestamp denotes the time when the data

UBICOMP/ISWC '15 ADJUNCT, SEPTEMBER 7–11, 2015, OSAKA, JAPAN

is obtained from the sensor module.

the trucks will periodically return to their car barns, e.g., after they finish the job of a day. It is envisioned that this opportunity can be employed as a supplement to the system for collecting large-volume data that tolerate a certain degree of delay.

Conclusion A public vehicle-based urban sensing system was proposed, implemented and experimented. In our future study, we will enlarge the experiment scale and test the scalability of the system. Adaptive data compression will be also investigated in order to improve the data collection efficiency. Figure 6: A photo after sensor node installation.

Preliminary Experiment and Discussion We conducted experiment to validate the applicability of the propose system. In the experiment, we installed 7 sensor nodes (one per group). By the time when this paper is written, the experiment has lasted for 4 weeks, conforming that the proposed system can work stably. The data are being published on [2] with the check box named FujisawaCarSensor1, 2, ..., 7. A photo after sensor node installation is shown in Figure 6. In the experiment, we made the following observations. First, strong temporal and spatial correlations exist in the collected data, implying that the communication efficiency of the system can be improved by adopting appropriated data compression algorithm. Second, the LTE network is impractical to collect large-volume data such as videos or images, due to the limitations and restrictions of the mobile network service.1 It is interesting to note that all 1 A typical restriction in Japan is 7GB per month for a user, beyond of which the link speed will be greatly reduced.

120

Acknowledgements This work was partly supported by Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan.

References [1] Burke, J. A., Estrin, D., Hansen, M., Parker, A., Ramanathan, N., Reddy, S., and Srivastava, M. B. Participatory Sensing. Center for Embedded Network Sensing (2006). [2] HT Lab. Simple SOX PubSub Demo. http://sox.ht.sfc.keio.ac.jp/tools/soxClient2.html, last accessed June, 30th, 2015. [3] Nakazawa, J., Takuro, Y., Ito, T., Ogawa, M., and Sakamura, M. Keio Universal Sensor Network System Manual based on Sensor-Over-XMPP. http://sox.ht.sfc.keio.ac.jp. [4] RASPBERRY PI FOUNDATION. http://www.raspberrypi.org. [5] The XMPP Standards Foundation. http://xmpp.org.