Progress of Wakkanai Experimental Community … over the course of five years. The report is...

1 Introduction

The Wakkanai Experimental CommunityNetwork Project is a collaborative researchproject that began in 1999 in Wakkanai City,Hokkaido, the northernmost city of Japan.This project was designed to study the techni-cal difficulties involved in the implementationof a wireless community network consistingof wireless equipment not requiring userlicenses, including low-power data transmis-sion systems (referred to below as “wirelessLANs”) and laser optical-beam networks. Theproject was also aimed at establishing a basisfor developing applications that make use ofthe advantages of broadband communicationin large, sparsely populated areas. This paperpresents the progress and achievements of thisproject over the course of five years. Thereport is constructed as follows: Section 2explains the background at the outset of theproject, Section 3 discusses the goals of theproject, Section 4 describes the five-yearprogress of the project, Section 5 providesdetails regarding the configuration of theExperimental Network, Section 6 summarizesthe achievements of the project, and Section 7

comments on the remaining challenges.

2 Background at the outset of theproject

Wakkanai Hokusei Gakuen College(referred to below as “Wakkanai Hokusei”)was established in 1987 as a junior college. Itis now a four-year college consisting of a sin-gle department: the Department of IntegratedMedia within the Faculty of Integrated Media.Since its establishment, the college has devot-ed particular effort to promoting information-processing education, including training pro-grams for UNIX system administrators. In1995, Wakkanai Hokusei established wirelessLAN connections to two prefectural highschools in Wakkanai (Hokkaido WakkanaiHigh School and Hokkaido Wakkanai ShokoHigh School). Wakkanai Hokusei thus veryearly on recognized the potential of applyingwireless technologies to a community net-work, and has since devised a number ofunique techniques in accomplishing this aim[1].

Meanwhile, the Communications ResearchLaboratory (the “CRL”) was searching for

TAKIZAWA Osamu and KANAYAMA Noriyo 193

Progress of Wakkanai ExperimentalCommunity Network ProjectTAKIZAWA Osamu and KANAYAMA Noriyo

Since 1999, CRL and Wakkanai Hokuseigakuen College have conducted a research proj-ect on implementation of wireless community network in Wakkanai, the northernmost city ofJapan. The city is suitable for experimental field of wireless self-supporting networkbecause of sparse population. However weather condition, i.e. high wind and heavy snow,is hard for wireless network. This paper introduces 5 years detailed progress of the project.

Keywords Community network, Wireless LAN, Optical beam network, Wakkanai city, Schoolinternet

194

new uses for the Wakkanai Radio Observatorylocated in the urban area of Wakkanai. TheCRL recognized that the implementation of acommunity network was a research topic thatcould be assessed even from remote locations.One of the authors (Takizawa), at the time amember of the Network and Information Sys-tems Office of the CRL Planning Division,proposed a collaborative research projectbetween Wakkanai Hokusei and the CRL.The two parties agreed to conclude a contractto form a field study on the implementation ofa community network mainly based on wire-less links, in a large, sparsely populated areaunder severe natural conditions. On October14, 1999, the President of the CRL visitedWakkanai City to deliver a lecture entitled“The Future of Information Media” to the gen-eral public of Wakkanai (Fig.1). The CRLthen began preparing equipment for the proj-ect at the Wakkanai Radio Observatory andother locations.

3 Goals of the project

Wireless LAN mainly uses ISM-band (2.4-GHz band) radio waves to connect personalcomputers within a local area, which couldextend up to several tens of meters, to the net-work infrastructure. However, communica-tion is possible between access points severalkilometers apart using wireless LAN equip-ment equipped with external antennas

approved under the Technical RegulationsConformity Certification protocol. Further,laser-beam network equipment allows forcommunication between locations up toapproximately one kilometer apart with bitrates ten to twenty times faster than IEEE802.11b standard wireless LANs (11 Mbps).This sort of equipment does not require userlicenses. Anyone can install these systemsand use them immediately.

To construct a fast network infrastructure,fiber-optic cables are the standard choice.However, new fiber optic equipment entailssignificant installation and maintenance coststhat increase with distance, including the costsof poles and ground rent. ADSL is a possiblealternative but performs poorly when the tar-get location is far from telephone stations.Thus, it is logical to conclude that connectionvia wireless systems—including wirelessLANs, laser optical-beam networks, and relaysystems—offers cost advantages when imple-menting a fast network infrastructure in large,sparsely populated areas. Accordingly, manyprojects are currently underway throughoutJapan involving the construction of communi-ty networks using wireless equipment. Oneexample may be found in the “ExperimentalDevelopment of a Rural Multimedia System”projects (for instance [2]) of the Telecommuni-cations Advancement Organization of Japan(the “TAO;” now integrated within the Nation-al Institute of Information and Communica-tions Technology, or “NICT”).

In terms of climate, Wakkanai City is notonly a cold, snowy area but it is also one ofthe windiest areas in Japan due its geographi-cal disposition facing the Soya Strait. Assuch, the natural environment is extremelysevere. The effects of the cold, snow, andwinds on antennas and radio conditions cannotbe ignored in a high-frequency wireless net-work using the 2.4-GHz band or lasers. Thus,the project established as one of its goals veri-fication of the practicability of establishing awireless network in such a severe natural envi-ronment. Many of the large, sparsely populat-ed areas in Japan are in Hokkaido, Tohoku,

Journal of the National Institute of Information and Communications Technology Vol.51 Nos.1/2 2004

Lecture by former Director Takashi Iidaof the CRL (At Wakkanai Hokusei,October 14, 1999.)

Fig.1

and Hokuriku, all with plentiful snowfall. Thefindings of this project were hoped to serve asa reference in promoting the acceleration ofcommunity network installation throughoutJapan.

The project was planned to progress as fol-lows. In the first phase, several organizationsin the city were to be connected with wirelessequipment, forming the “Wakkanai Experi-mental Community Network” (referred tobelow as “the Experimental Network*”). Thesecond phase was to involve the developmentof applications to run on the constructedExperimental Network. The organizations tobe connected to the Experimental Networkwere to have public use and would also func-tion as test beds for application development.With these factors in mind, junior and seniorhigh schools in Wakkanai were selected as tar-get organizations.

The Experimental Network was construct-ed as shown in Fig.2 as result of this collabo-rative research. Wakkanai Hokusei serves as

the main hub station, connecting three seniorhigh schools, five junior high schools, andWakkanai Radio Observatory (unmannedoperations beginning in June 2003).

The purpose of the project was not simplyto construct a community network. The inten-tion was to facilitate advanced and practicalexperimental research in view of next-genera-tion IT technologies (such as IPv 6) and toestablish flexible designs enabling dynamicresponse to the sudden alteration of pathwaysthat are inevitable within a wireless network.Care was also taken to ensure that the con-nected schools would have a sense of activeparticipation in the experiment from the startof network construction, as opposed to passiveuse of a given community network. Toachieve this aim, Wakkanai Hokusei providednetwork administration training to teachers inthe connected schools, as described in Section5. These sorts of activities in the implementa-tion of a community network contribute sig-nificantly to the improvement of IT skills in


* Those involved in the project generally referred to the Experimental Network as the “3K Net,” the three Ksstanding for “Koiki (large area),” “Kaso (sparsely populated),” and “Kakoku-na shizen-kankyo (severe naturalenvironment).”

Connection diagram of the Wakkanai Experimental Community Network (March 2004)Fig.2

196

primary and secondary education. One of theauthors (Takizawa) co-edited a textbook totrain “Joho” teachers, a new resource forinformation technology education in seniorhigh schools [3], leading to the inclusion of anumber of comments on this project.

4 Progress of the project

4.1 Progress in 20004.1.1 Reinforcement within the

Wakkanai Hokusei organizationIn April 2000, directly after the start of the

project, Wakkanai Hokusei became a four-year college, leading to improvement in thespecialized skills of the students who partici-pated in the project.4.1.2 New connections

On June 29, 2000, the Soya District Officeof Education, a local agency of the HokkaidoPrefectural Board of Education, designatedtwo prefectural senior high schools, namely,Wakkanai High School and Wakkanai ShokoHigh School, as project participant schools.This designation clarified the roles within theproject of these two schools, which had beenconnected with wireless LANs since 1995through Wakkanai Hokusei’s independent net-work activities. The collaboration of theWakkanai City Board of Education was alsosecured under the project, clearing the way forconnection of the junior high schools inWakkanai to the network.

After these preparations, the project pro-ceeded to connect each of the city junior highschools to Wakkanai Radio Observatory andthe network via wireless LAN.(1) Between Wakkanai Radio Observatory and

Wakkanai Hokusei Wakkanai Radio Observatory is located in

a low-lying depression out of direct view ofWakkanai Hokusei. Wakkanai Otani HighSchool, a private high school, is situatedbetween these two locations, and it becameclear that relaying via the roof of the gymnasi-um at the high school would allow for com-munication between the observatory andWakkanai Hokusei. With the cooperation of

Wakkanai Otani High School, a set of relayantennas was installed to this end. Figure 3shows the relay antennas installed on the gym-nasium of the high school. The distancebetween Wakkanai Radio Observatory andWakkanai Otani High School is approximately640 m, and the distance between the highschool gymnasium and Wakkanai Hokusei isapproximately 2.9 km. Yagi antennas wereused at the high school and college connected.The antenna at Wakkanai Radio Observatorywas installed indoors, near a window on thesecond floor of the office building. Figure 4shows the Yagi antenna installed outdoors atWakkanai Hokusei, pointing toward WakkanaiOtani High School. At this stage, only the


Relay antennas installed on the roof ofthe gymnasium of Wakkanai OtaniHigh School. The antenna on the left is for connec-tion with Wakkanai Hokusei and theantenna on the right is for connectionwith the Wakkanai Radio Observatory.

Fig.3

Yagi antenna installed outdoors forthe wireless LAN to Wakkanai OtaniHigh School (Wakkanai Hokusei)

Fig.4

relay antennas were installed at WakkanaiOtani High School, without implementing thenetwork in the school. Due to CRL securitypolicy, the Experimental Network was notconnected to the CRL’s own LAN.

Because the Experimental Networkextended to the Wakkanai Radio Observatory,a web server was installed at the observatoryspecifically to present general aspects of theproject, with a portal site opened to the publicin March 2001 (http://www.crl.wakkan-ai.ne.jp/). (2) Between Wakkanai Higashi Junior High

School and Wakkanai Shoko High SchoolThe wireless LAN was extended to

Wakkanai Higashi Junior High School,approximately 300 m from Wakkanai ShokoHigh School, with the latter already connectedto the wireless network extending fromWakkanai Hokusei. The Wakkanai ShokoHigh School placed an antenna in its observa-tory and Wakkanai Higashi Junior HighSchool placed a planar antenna near a windowin a classroom. Figure 5 shows the antennainstalled at Wakkanai Higashi Junior HighSchool, facing Wakkanai Shoko High School.

(3) Between Wakkanai Minami Junior HighSchool and Wakkanai High SchoolAs Wakkanai Minami Junior High School

is located just behind the Wakkanai RadioObservatory, it seemed desirable to connectthis high school to the wireless LAN via the

Wakkanai Radio Observatory. However, forconvenience in network management, the jun-ior high school was connected to the LAN viaWakkanai High School, which was alreadyconnected via Wakkanai Hokusei. A planarantenna was selected for the Wakkanai Mina-mi Junior High School, installed indoors neara classroom window.(4) Between Shiomigaoka Junior High School

and Wakkanai Hokusei Shiomigaoka Junior High School is locat-

ed directly adjacent to the grounds ofWakkanai Hokusei, with no third-party prop-erty between the two. Based on calculationsshowing that it would be less costly than pur-chasing and installing wireless network equip-ment, fiber-optic cables were installed onpoles situated between the two organizations,providing connection at a transmission rate of100 Mbps. Figure 6 shows the cables con-necting the two organizations.

During this period, a “Community Net-work Committee” was organized, includingthe Director of the Wakkanai Radio Observa-tory, representatives from Wakkanai Hokusei,Planning and Coordination Division, Munici-pality of Wakkanai, the Wakkanai City Boardof Education, and headmasters of the connect-ed schools, to conduct various negotiationsbetween the connected organizations. Thecommittee held three meetings: on August 22,2000, December 18, 2000, and March 21,


Planar antenna installed indoors forthe wireless LAN to Wakkanai ShokoHigh School (Wakkanai Higashi JuniorHigh School)

Fig.5

Fiber-optic cables connecting Wak-kanai Hokusei (center) and Shiomi-gaoka Junior High School (to the leftbeyond the frame of the photograph)

Fig.6

198

2001.As the number of connected organizations

increased, Wakkanai High School andWakkanai Shoko High School started to playthe roles of hub stations, which demandedreinforcement in the links to Wakkanai Hoku-sei. Thus, the wireless LAN equipment (1.5Mbps) installed by Wakkanai Hokusei in 1995was replaced by laser-beam network equip-ment, both between Wakkanai Hokusei andWakkanai High School and between WakkanaiHokusei and Wakkanai Shoko High School.Figures 7 and 8 show the laser-beam networkequipment installed at Wakkanai Hokusei.The CRT display shown in the photograph isthe monitor used to adjust the laser beam.

4.1.3 Providing services open to thepublic

In June 2000, the All-Hokkaido High

School Basketball Championship was held atseveral locations, including Wakkanai ShokoHigh School. As a demonstration of theExperimental Network, the games were cov-ered live on the Internet, an achievementreported in the local newspaper [4].

4.2 Progress in 20014.2.1 Beginning to provide information

As the Experimental Network was operat-ing smoothly, Wakkanai Higashi Junior HighSchool and Shiomigaoka Junior High Schoolestablished websites in March 2001 and beganuploading information introducing theirschools to the Internet; these activities werealso reported in the local newspaper [5].4.2.2 Clarification of the CRL organiza-

tionIn April 2001, the CRL was reorganized as

an incorporated administrative agency, and theradio observatory was eliminated from theorganization. The Emergency Communica-tions Group of the CRL Information and Net-work Systems Division thus determined totake charge of this project on behalf of theCRL, with the establishment of a clear organi-zation to promote the project. At the sametime, the project schedule was fixed: theexperiment was slated to end at the end ofMarch 2003, and after one year of follow-up,collaborative research was to end in March2004, five years from the start of the project.4.2.3 Observation of correlation

between network performanceand meteorological data

To verify the practicability of a wirelessnetwork under cold, snowy, and windy condi-tions, it is important to study the general cor-relation between meteorological and networkconditions. Thus, to observe the correlationbetween network performance and meteoro-logical data provided by the Wakkanai LocalMeteorological Observatory, in November2001 the project began to use Meteor i-NET,an Internet data service provided by the JapanMeteorological Business Support Center(JMBSC). By connecting to a server at theJMBSC, a user can download quasi-real-time


Laser-beam network equipment forconnection to Wakkanai High School(Wakkanai Hokusei)

Fig.7

Laser-beam network equipment forconnection to Wakkanai Shoko HighSchool (Wakkanai Hokusei)

Fig.8

meteorological data, including surface obser-vations and AMeDAS data. The results ofthese correlation observations are discussed in6.2.1 in detail.4.2.4 Extension to Wakkanai Junior

High SchoolIn August 2001, Wakkanai Junior High

School was newly connected to WakkanaiHigh School by wireless LAN. WakkanaiJunior High School is the only school amongthe connected organizations located in thecentral section of Wakkanai City and is out ofdirect view of Wakkanai High School. Thus,relay equipment was installed at WakkanaiCity Hall and at the City Library. Only relayequipment was used; no network was installedat these two locations.4.2.5 Appeal and reputation of the

projectAround this time, the project started to

attract public attention for its grass-rootsimplementation of a community network withwireless equipment. Accordingly, massmedia, such as local television news, beganreporting on the project with increasing fre-quency [6] [7]. Wakkanai Hokusei was alsoawarded the Hokkaido TelecommunicationConference Chairman’s Award during theInformation and Communications Month(May 15–June 15), 2001, for its contributionsin the implementation of the ExperimentalNetwork.

4.3 Progress in 20024.3.1 Extension to Wakkanai Otani

High SchoolIn March 2002, Wakkanai Otani High

School was connected to the ExperimentalNetwork. The link was branched via the relayequipment already installed on the schoolgymnasium to connect Wakkanai RadioObservatory and Wakkanai Hokusei. Connec-tion to a school building approximately 100 maway was provided by wireless LAN.4.3.2 Countermeasures against snow-

driftCorrelation observations in relation to

meteorological data that began in fall 2001

showed that the link between Wakkanai Hoku-sei and Wakkanai High School, which wasprovided by a laser-beam network system overa particularly long distance (1.9 km), was fre-quently disconnected due to snowdrift andsimilar causes. Thus, beginning March 2002,the connection for Wakkanai Hokusei,Wakkanai High School, and Wakkanai ShokoHigh School was triangularly duplexed vialaser-beam network and wireless LAN. Thisprovided means to bypass the connectionroute dynamically using OSPF in the case ofsnowdrift. Figure 9 shows the two sets oflaser-beam network equipment (each pointingtoward Wakkanai Hokusei and WakkanaiShoko High School) and two wireless LANantennas for the backup link installed inWakkanai High School. Details are discussedin 6.2.

4.3.3 Workshop on the collaborationresearch

From July 2 through July 4, 2002, projectparticipants from the CRL and WakkanaiHokusei held a “Collaboration ResearchWorkshop” at Wakkanai Hokusei (Fig.10).Eight people participated in the workshop:from the Emergency Communications Groupof the CRL Information and Network SystemsDivision, the CRL Planning Division, andWakkanai Radio Observatory. The workshopincluded presentations and discussions by stu-


Duplex network equipment installed atWakkanai High School: two sets oflaser beam network equipment (forconnection to Wakkanai Hokusei andWakkanai Shoko High School) and twowireless LAN antennas for the backuplink

Fig.9

200

dents from Wakkanai Hokusei, a tour of theconnected organizations, and discussions withthe Planning and Coordination Division,Municipality of Wakkanai.

A large fire had taken place in the centralpart of Wakkanai City, involving the totaldestruction of 23 buildings on June 29, justbefore the opening of the workshop. The par-ticipants also visited the site of the fire.4.3.4 Presentation of the project on

CRL open-house daysOn August 2 and 3, 2002, the CRL held

open-house days at its headquarters inKoganei City, Tokyo, with the activities of theproject displayed in the exhibition, with manyattendees showing interest in the prospect oflong distance communications based on wire-less LANs (Fig.11).

4.4 Progress in 20034.4.1 Information and Communica-

tions Symposium in WakkanaiOn May 15, 2003, the “Information and

Communications Symposium in Wakkanai”was held, cosponsored by the HokkaidoBureau of Telecommunications, the Informa-tion and Communications Month PromotionCouncil, the Hokkaido TelecommunicationConference, and Wakkanai City Government.One of the authors (Kanayama) served as apanelist, introducing the project’s activities.On May 16, a videoconference experimentwas performed between Shiomigaoka JuniorHigh School and Wakkanai Junior HighSchool. However, due to malfunction of thePC router kernel and the implementation ofmulticasting, which will be discussed in 6.2.2,it was necessary to use IP tunneling with theDVMRP. 4.4.2 Extension to Soya Junior High

SchoolIn spring 2002, regulations concerning

low-power data communication systems wereeased, allowing for antenna gain of up to12.14 dBi [8]. As a result, nominal 16-kmproducts became available without the needfor a user license. Thus, In July 2003, projectmembers attempted a direct connection over adistance of approximately 17 km betweenSoya Junior High School and Wakkanai ShokoHigh School (Fig.12). From the Wakkanaiurban area in the direction of Cape Noshapputo Soya Junior High School in the direction ofCape Soya, a connection was established overthe Soya Bay with no problems. Soya JuniorHigh School is the northernmost junior highschool in Japan, located just before CapeSoya. In 2003, the network infrastructurecommercially available to the public in thelocal area of the school had been limited to64-kbps ISDN services. This link thus estab-lished a permanent connection to and from thenorthernmost part of Japan at a transmissionrate enabling conveyance of video. Livevideo taken by the network camera installed inthe school’s observatory were continuouslyexhibited (Fig.13); again, these achievements


Collaboration Research Workshop(from July 2, 2002 through July 4,2002, at Wakkanai Hokusei)

Fig.10

Display at the open-house exhibition(August 2 and 3, 2002, CRL Head-quarters)

Fig.11

were documented in a related trade paper [9]. During the CRL open-house days on

August 1 and 2, 2003, a well-received demon-stration took place in which visitors couldoperate the network camera at Soya JuniorHigh School from CRL Headquarters viaInternet (Fig.14).

The connection to Soya Junior HighSchool completed the first phase of the experi-ment—i.e., the implementation of the Experi-mental Network.

5 Configuration of the Experimen-tal Network

Each organization (node) connected to theExperimental Network is treated as a sub-domain in the Wakkanai Hokusei campusLAN (wakhok.ac.jp). In May 2001, WakkanaiHokusei changed the Experimental Networkpart of the LAN to a double-domain networkand the nodes were assigned to apparentlyindependent domains expressed asXXX.wakkanai.ne.jp as viewed from theInternet. (Here, XXX indicates a sub-domainname that varies from one node to another.)All nodes are located within the firewall of theWakkanai Hokusei LAN except for the respec-tive web servers, the settings for which aremade to allow access from external networksby port forwarding. Each node can be reachedin accordance with application of theWakkanai Hokusei policy regarding globaladdresses. One of the authors (Kanayama) isalso one of Wakkanai Hokusei’s networkadministrators, and thus was in a position to


Parabolic antenna installed at SoyaJunior High School for long distancewireless LAN

Fig.12

Screen shot of video images takenby the network camera installed inthe observatory of Soya Junior HighSchool (Cape Soya at upper-rightcorner)

Fig.13

Display at the open-house exhibition (August 1 and 2, 2003, CRL Headquarters)Fig.14

202

design the Experimental Network out ofWakkanai Hokusei’s LAN such that these twonetworks were treated as equivalent. Howev-er, it was decided that the Experimental Net-work should temporarily remain within theWakkanai Hokusei LAN until each node couldindependently handle the functions of networkmanagement. Today, when the nodes demandmore services, the Experimental Network canrespond flexibly.

Each node is assigned to an independentsub-domain. Wakkanai High School andWakkanai Shoko High School are hub sta-tions, each with several personal computers(PCs) constituting a network. Other organiza-tions (junior high schools and Wakkanai RadioObservatory) are terminal nodes, each with asingle PC equipped with network server func-tions. At least single global IP address, withnetmasks of 28 bits, is assigned to each PC.For routing purposes, the Experimental Net-work is composed of private addresses aswell. The private addresses of the Experimen-tal Network are designed to maintain consis-tency with those of the Wakkanai Hokuseiintranet. (Wakkanai Hokusei’s private addressis 10.0.0.0/8, while the Experimental Net-work’s private address is 10.17.0.0/16.) Thisdesign will allow for extremely convenientmodification of the network in the future,when the nodes of the Experimental Networkand Wakkanai Hokusei become treated asequivalent network components. When BFlets service of NTT becomes available inWakkanai, this design will also allow a VPN(Virtual Private Network) connection betweenthe nodes and Wakkanai Hokusei using BFlets, and it will also be easy to control rout-ing between the wireless network and B Flets.In any case, the gateway PCs of the Experi-mental Network are currently placed withinthe Wakkanai Hokusei intranet routing. Thus,viewed from the Internet, the network featuresdouble-wall structures with independent fire-walls for the connected organizations based ontheir own policies, within the firewall of theWakkanai Hokusei campus LAN.

Figure 15 represents a selected section of

the Wakkanai Hokusei network structure as itrelates to the Experimental Network.

Ordinary PCs equipped with the FreeBSDOS are used for all routing in the Experimen-tal Network. Of course, special-purposerouters could also be used. However, moreflexible routing using PCs was selected, inconsideration of the possibility of performingexperiments into new routing methods andmulticast routing. Fortunately, branchingwithin the Experimental Network can be han-dled by the PCs. Nevertheless, it should benoted that the actual installation conditionsrequire a variety of techniques: for example,low-band wireless LANs are handled within asingle line in the hub stations based on VLANtechnology allowing for the presence of twonetworks within structures that are physicallythe same but logically different. For this rea-son, an L2 switch that can handle VLAN andan exclusive routing PC are placed at each ofthe two hub stations (Wakkanai High Schooland Wakkanai Shoko High School). The PCitself handles the VLAN tag. Figure 16 shows


Schematic diagram showing theconnection between WakkanaiHokusei LAN and the ExperimentalNetwork

Fig.15

Schematic diagram of the internalnetwork within Wakkanai High School

Fig.16

the network at Wakkanai High School, one ofthe hub stations.

Due to its dependence on laser and otherwireless equipment (as discussed later), apractical problem in the Experimental Net-work arises in maintaining the links betweenthe hub stations at Wakkanai High School,Wakkanai Shoko High School, and WakkanaiHokusei. To solve this problem, project mem-bers settled on the use of wireless LANs andredundant connection to other schools. Thus,double lines connect these three organizations,as explained in 6.2.2. Due to the cable rout-ing in each of the schools and other reasons,equipment details differ among the organiza-tions—including line concentration and theVLAN relay—but the logical structure is basi-cally the same at Wakkanai Shoko HighSchool and at the remaining stations. Theonly difference is that Wakkanai Shoko HighSchool is connected to Soya Junior HighSchool 17 km away, and linked to WakkanaiHigashi Junior High School within a closerange, via wireless LAN.

The basic design is the same for all nodes.In the junior high schools, a single PC pro-vides the functions of a firewall, an NAT box,a mail server, a web server, and a DHCP serv-er. Administration of these servers is in theorythe responsibility of each node, but WakkanaiHokusei was at first delegated with this role.Wakkanai High School and Wakkanai ShokoHigh School participated in a collaborativeconnection experiment set up by WakkanaiHokusei before the start of the project. Thus,Wakkanai Hokusei took responsibility foradministration at first, but now the highschools themselves are responsible for admin-istration. As a matter of course, WakkanaiHokusei has provided assistance to these highschools, including training for administrators.The equipment was also gradually upgradedand expanded from a configuration similar tothose of the junior high schools describedabove to a more extended configuration withseparate server functions. The lessons of thisexperience led to the application of the sameapproach to the junior high schools. Some of

these junior high schools are now, initiatingindependent administration. In this sense,another significant effect of the ExperimentalNetwork can be seen in the growth withineach organization, both in terms of humanresources and in terms of network equipment.

Another characteristic of the node designis that the server PC is always installed in thestaff room or in the principal’s office to con-trol the existing internal network of each node.As the two high schools joined the networkearlier, they have grown internally since theinstallation of the servers, improving theirinternal networks. However, the junior highschools already had PC rooms or staff-roomnetworks when they were connected to theExperimental Network. These networksmight have been insufficient in terms of exter-nal connections and each had a different con-figuration. Nevertheless, each school hadsome form of existing network. It has beensuggested that one of the reasons behind thesuccess of the project is that the ExperimentalNetwork was planned to integrate (or simplyproved capable of integrating) these internalnetworks, as opposed to being designed as anexternal entity. Of course, it was fortunatethat the project received the full cooperationof the City Board of Education. In any case,as discussed above, the node for each juniorhigh school was designed as shown in Fig.17.

As Soya Junior High School, the most dis-tant of all sites, already had ISDN networkfacilities before connection to the Experimen-tal Network, and as the stability of the wire-less link was unknown, the firewall settingswere adjusted to switch the network automati-


Schematic diagram of the internalnetwork within each of the juniorhigh schools

Fig.17

204

cally to Flets ISDN connection in the event offailure in the wireless link.

6 Achievements

6.1 Comparison with similar communi-ty networks

6.1.1 Kamifukuoka Wireless LAN Pro-ject

The KDD Research and DevelopmentLaboratories (now KDDI R&D Laboratories)hosted an experimental project conducted inKamifukuoka City of Saitama Prefecture fromDecember 1999 through to July 2000. Elevenschools and ten to twenty individuals partici-pated in this project. The experiment alsoincluded multicast broadcasting. However,this was a simple broadcast experiment with-out actual multiple routing due to the experi-mental configuration, with the wireless equip-ment installed on a central steel tower. Theexperiment was also limited in terms of therange of communication to and from the steeltower, a range constituting a circular area witha diameter of approximately 3 km.6.1.2 Community Network in the town

of Wajiki, Tokushima PrefectureThe Shikoku Bureau of Telecommunica-

tions hosted a study group investigating theuse of small-scale wireless systems for com-munity networks, and this group subsequentlyconducted an applied experiment. This was arelatively early experiment and is thus consid-ered a pioneer study. ROOT Inc. developedthe equipment used in the experiment. At theoutset, the equipment was based on NetBSD,although it has now evolved to form an exclu-sive unit (the content appears still to be basedon NetBSD). As a result of this development,the equipment can in theory handle wirelessrouting. On the other hand, the unit is com-mitted to certain preset purposes and it wouldbe difficult to compare various routing algo-rithms and to introduce multicast routing.Nevertheless, it stands as a tried-and-testedspecial-purpose unit and has been introducedin different areas in Japan (including the TaiseiTown in Hokkaido, Aomori City, and Mat-

suyama City). This unit is also used in thecurrent project for the connection betweenWakkanai Shoko High School and Soya JuniorHigh School.6.1.3 Maebashi City Educational Infor-

mation Network The Maebashi City Educational Informa-

tion Network began as a service based on theAdvanced Educational Network Model Pro-ject (supported by the former Ministry of Edu-cation, Science, Sports and Culture and theMinistry of Posts and Telecommunications) in1999. Network connection extended to 20schools, (elementary, junior high, and seniorhigh), as well as to the Children’s Center, with1.5-Mbps wireless links, and to 12 additionalelementary schools and one other organizationvia a 0.5-Mbps link. Approximately 30 organ-izations continue to be involved, with elemen-tary and junior high schools as the main par-ticipants. One particular characteristic of theendeavor is that the “Gunma Internet Volun-teers for Education” (GIVE), a volunteerbased organization, provides most of the net-work design and support. This project is simi-lar to the current project in many aspects.However, research in the former case is stillmainly for the purpose of education.6.1.4 Advanced Educational Network

Model Project As discussed above, Maebashi City estab-

lished a network based on this model project,supported at inception by the former Ministryof Education, Science, Sports and Culture andthe Ministry of Posts and Telecommunica-tions. Other examples of independent networkimplementation based on this project includeShizuoka City, Hamamatsu City, and SetagayaWard in Tokyo, in all of which 10 to 20schools were connected via laser (with physi-cal speeds of 155 Mbps). Other exampleshave involved the use of satellite communica-tions. All 30 models of this example employdifferent methods.6.1.5 Naruto University of Education

Naruto University of Education is inde-pendently connected to two local junior highschools via laser (100 Mbps), but the details of


this system are not currently available, includ-ing the purpose of the project.

6.2 Experiments and discussion oflaser and wireless LAN networks

6.2.1 Meteorological conditions andconnection ratio

From the start of the project, snowfall wasexpected to influence the laser connections.In a multicast experiment performed in thefirst spring, dense fog rendered communica-tion impossible for a time, which raised con-cerns about the issue of snowfall. Figure 18plots the daily connection ratio from February28, 2001 to March 21, 2001. Here the connec-tion ratio is defined as the ratio of reply pack-ets to sent packets when five 64-byte ICMPpackets are transmitted per minute. As shownin the Figure, the connection conspicuouslyfailed on March 4 and March 7.

Figures 19 and 20 plot the weather condi-tions and the connection ratio on these twodays with the same time axis. The weatherdata plotted include wind speed, range of visi-bility, and snowfall observed at the WakkanaiLocal Meteorological Observatory. Unfortu-nately, the observatory is near Wakkanai CityHall, which is approximately four kilometersfrom the observation point for measuring the

connection ratio (Fig.2). Wind speed has asignificant effect on conditions in a snow-storm, specifically in terms of the range of vis-ibility. This is due to the fact that the snow isconsistently powdery, with ground tempera-tures of approximately -10℃, after the begin-ning of winter and before early spring. Thisrenders the effect of snowfall on the range ofvisibility largely dependent on the presence orabsence of wind. Even when there is nosnowfall, snow accumulated on the groundsometimes acts like falling snow as the winds


Connection ratios of the laser beamnetwork

Fig.18

Connection ratios and meteorologi-cal conditions (March 4, 2001)

Fig.19

Connection ratios and meteorologi-cal conditions (March 7, 2001)

Fig.20

206

stir up the loose snow. This phenomenon isreferred to as snowdrift. Thus, snowfall andwind speed were considered important weath-er factors at the start of the experiment, andthe correlation between these weather condi-tions and the connection ratio was taken intoconsideration. It should be noted that theweather data obtained from the Meteor i-NETservice was provided in a unique format. Areading tool was provided; nevertheless, it wasextremely inconvenient and a new tool wasdeveloped for use in the project. In Fig.18,the connection ratio for communication toWakkanai High School is indicated in red andthe ratio for Wakkanai Shoko High School isshown in black.

As previously indicated in Fig.2, while thedistance between Wakkanai Hokusei andWakkanai Shoko High School is approximate-ly one kilometer, the distance betweenWakkanai Hokusei and Wakkanai High Schoolis twice as far, at approximately two kilome-ters. Considering that energy loss for wirelessLANs or laser transmission is proportional tothe square of distance and that direct visibilityis proportional to distance (assuming constantsnowfall), the difficulty in connection betweenWakkanai Hokusei and Wakkanai High Schoolis estimated to be over three times that ofcommunication between Wakkanai Hokuseiand Wakkanai Shoko High school. Figure 19shows that the connection ratio dropped rapid-ly before 13:00, recovered after 17:00 for atime, but deteriorated again. Here note thatsnowfall at 15:00 was 0 mm according to theweather data, but this is probably an anomalyrelating to the location of the observationpoint; a certain extent of snowfall and simulta-neous wind were considered to be the causesof disconnection at this stage. Next, Fig.20clearly shows the difference between condi-tions at Wakkanai High School and WakkanaiShoko High School. The weather data at 9:00shows that there was a considerable amount ofsnowfall from morning to noon, but it appearsthat snowfall generally decreased later, asshown in the accumulation of 10 mm between9:00 and 15:00. Reflecting these weather con-

ditions, the connection to Wakkanai ShokoHigh School recovered to approximately 50 %of full performance, while the connection toWakkanai High School remained at 0 %.Even more interestingly, the connection gener-ally recovered between 15:00 and 21:00 inspite of the fact that snow accumulation dur-ing this period was the same as in the previousobservation. These results indicate that snow-fall had already decreased to a low level ataround 15:00 in the afternoon but that snow-drift occurred as a result of the morning snowand winds of approximately 10 m/s. Therange of visibility data also reflects this state:the range was substantially below one kilome-ter but then recovered to approximately onekilometer by 21:00.

Based on this data, the laser connectionsbetween Wakkanai High School, WakkanaiShoko High School, and Wakkanai Hokuseiwere improved to allow for combined use withwireless LAN. Figure 21 shows the weatherdata and connection ratio measured on oneday the following year, with the improved net-work. A constant snowfall of approximately10 mm was observed all day, but the figureclearly shows that the connection ratio for thelaser link varied due to the wind. On the otherhand, no significant changes were noted in thewireless LAN connection in spite of theseconditions. Weather factors such as snowfallalso influence wireless LANs to a certainextent; however, the conditions surroundingthe communication equipment influence theconnection far more than snowfall or rainfallbetween the two locations connected. Forexample, in the Wakkanai Hokusei experimentperformed before the current project, the wetsnow in the early spring covered the front sideof the planar antenna and completely blockedthe connection. Similar problems alsooccurred in this experiment. As planar anten-nas often present these sorts of problems, theywere always installed indoors in the earlystage of the experiment, with connectionestablished through a window. Nevertheless,at Wakkanai Higashi Junior High School, thewindow was covered with snow, which


blocked the connection. This adhesion ofsnow does not often occur in heated roomswith human activity. However, at HigashiJunior High School, the equipment wasinstalled in an infrequently used room due tophysical restrictions stemming from the con-nection to the hub station at Wakkanai ShokoHigh School. Later, the main antenna path forthe wireless LAN was changed, with a switchfrom planar antennas to Yagi antennas; withthese changes these interruptions seldomoccur, even when the antennas are installedoutdoors. This is the reason for the apparentsynchronization of the data for the wirelessLAN connection between Wakkanai HigashiJunior High School and Wakkanai Shoko HighSchool and the data for the laser connectionbetween Wakkanai Hokusei and WakkanaiShoko High School. In any case, the datareveals that the wireless LAN is relativelyresilient, even with snowfall, and that no prob-lem arises if the connection route is switchedto wireless LAN when the laser connectionfails.

A technical problem remained in switchingthe connection from laser to wireless LANsafely and quickly. To effect this automaticswitching, dynamic routing with OSPF (OpenShortest Path First) protocol was introduced inthe connections between Wakkanai Shoko

High School, Wakkanai High School, andWakkanai Hokusei, as described in 4.3.2.However, OSPF protocol also changes theroute after detecting a period of disconnection.Consequently, the disconnection must bequickly detected, but there was a concern thatsuch a quick detection system would misiden-tify a temporary recovery as an ongoing con-nection, leading to detection errors. An initialexperiment was therefore performed to short-en the detection time. However, the PC in usewas not capable of handling these settings,and it became evident that the OS would hangwhen these settings were forced. This verifi-cation was thus unfortunately abandoned.

As discussed, there are various problemsin establishing countermeasures for winter-related challenges. Nevertheless, it is impor-tant to note that a fast network has been main-tained generally well throughout the year.There is no other example in Japan of success-ful stable operation of a network that supple-ments the weaknesses of laser connection witha practical network using inexpensive equip-ment such as wireless LANs. Figure 22 showsthe connection ratio during three months in thespring of 2002. The figure clearly shows thatthe laser connection also operates relativelystably around April, in contrast to February orMarch, when blizzards are frequent.


Connection ratios between various sites and weather dataFig.21

208

6.2.2 Duplex design with laser andwireless LAN

As discussed in 4.3.2 and 6.2.1, the con-nection routes were duplexed with laser andwireless LAN to address winter-related prob-lems in laser communications. The significantpoints of this duplex design were as follows.

The three sites to be duplexed wereWakkanai Hokusei, Wakkanai Shoko HighSchool, and Wakkanai High School. Takingthe maximum bandwidth between these sitesinto consideration, it was estimated that 200-Mbps communication would be required ateach site with full operability of the laser con-nection. In view of various technical experi-ments envisioned with this Experimental Net-work—including multicast and IPv6 applica-tions—equipment such as the dedicated L3switch would obviously be insufficient. It wastherefore determined to introduce routing viaPCs as in the connection points of the variousnodes. At the same time, in consideration ofthe bandwidth requirements and also in lightof heavy multicast use (IPv4 and IPv6), thelines would have to be separated; though thiswas not completely feasible due to the limita-tions of the node equipment, at least a logicalseparation was necessary. As a result, a routerPC equipped with a PCI-X bus and GigabitNIC was selected to secure bandwidth, and anL2 switch with a Tagging VLAN function wasincluded in the design to ensure separation.Figure 23 shows a conceptual diagram of the

design. Of course, the PC router must be ableto control Gigabit Ether and it must also beable to handle VLAN directly. Fortunately,BSD was updated at this time to supportVLAN, with operation confirmed in precedingexperiments. However, the OSPF and multi-cast functions were not sufficiently supportedat first—it was not until near the end of theproject that the duplex lines began to operateas designed.6.2.3 Remaining challenges

The remaining challenges include a num-ber of problems arising at the start of the proj-ect that were known from similar experiments,and other problems that still remain unad-dressed. This section outlines the problemsthat could not be studied during the course ofthe project.

As the authors were not specialists inmeteorology, the study of the correlationbetween the meteorological data and the con-nection ratio was in a sense insufficient. Inparticular, few existing studies are availableconcerning various snowfall-related problemsin wireless LAN or laser communications, andthus collaboration with specialists is indispen-sable. When one of the authors (Kanayama)delivered a lecture at Wakkanai Local Meteo-rological Observatory, a meteorologist offereda number of suggestions, which unfortunatelycould not be incorporated into this study.Specifically, we had limited our focus tosnowfall, but snowfall was measured every six


Connection ratios from February 21, 2002 through to May 21, 2002Fig.22

hours, which was not sufficiently frequent forcomparison of the data with the connectionratio. On the other hand, rainfall was meas-ured quite precisely, with some snowfall dataincluded in the rainfall data. Of course, thisled to the problem of distinguishing betweenrainfall and snowfall, but the meteorologistmentioned above pointed out that actual earth-surface temperatures could be used to makethis distinction. Hence it may be possible todraw new conclusions about the connectionratio based on the three-fold correlationamong rainfall, earth-surface temperature, andwind speed. We have not performed theseanalyses but data is plentiful and we hope tostudy this problem in the future.

A second problem was found in the duplexdesign, which unexpectedly hampered net-work operation with respect to the OS kernel,as noted earlier. The authors’ assessments inthis respect might have been too optimistic.Nevertheless, the network has at last begun tooperate stably, and multicast and other experi-ments can now be performed. Connection viaPC router has advantages in terms of flexibili-

ty, but stability in this case inevitably pales incomparison with that of a dedicated router.The current design was selected in accordancewith the purposes of the experiment, but itcannot be denied that this decision caused acertain degree of inconvenience among theparticipants.

6.3 Low-latitude auroral observationWhen the project began in 1999, the num-

ber of sunspots was at a maximum. Accord-ingly, continuous image monitoring of low-latitude auroral development was proposed. Anetwork camera was installed at WakkanaiHokusei for automatic imaging of the northsky in the direction of the Soya Strait at con-stant intervals at night, with data stored withinthe server. Basic tests and operational experi-ments were thus performed in the provision ofthis data to the public over the Internet.

6.4 Experiments at Wakkanai RadioObservatory

With the rapid growth in email using PCsand portable phones, a so-called “digital


Duplex designFig.23

210

divide” developed between those who werecommunicating via email and those who werenot. In response, to provide an easy means forthose uncomfortable with information-tech-nology equipment to enter into the world ofemail communication, an experimental email-to-voice conversion system was installed atthe Wakkanai Radio Observatory as an appli-cation experiment over the Experimental Net-work.

This system is based on “Mail Carrier,”developed by Railway Information SystemsCo., Ltd. for groupware mail systems, modi-fied to operate using SMTP. The system wasinstalled on an Internet mail server construct-ed in Windows NT, providing functions toread text aloud through voice synthesis and toidentify users based on caller ID and DTMF(touch-tone) signals. In this system, the usercalls in to listen to the voice-synthesized con-tent of emails delivered to the user’s mailserver account (a pull-type service). When anemail arrives with a certain flag, the experi-mental system immediately calls the targetuser; the user can then listen to the content ofthe email (a push-type service). Registereduser telephone numbers may also be modifiedby phone.

This system differs from the syntheticvoice functions implemented in client e-mailsoftware for the visually impaired in that herethe voice function is implemented on the mailserver. The system offers an advantage in thatthe user does not need to have access to hard-ware such as a PC or even a specific portablephone, but instead can listen to the email bycalling the server from a home or public tele-phone.

We prepared this system and solicitedapplications during the Wakkanai RadioObservatory open-house days in the summerof 2001. However, not a single personapplied. Meanwhile, portable phone e-mailservices, including those provided by i-mode,were rapidly spreading throughout society,nearly resolving the digital divide issue, whichhad been the motivation for the developmentof this system.

A server-side service such as this one isconsidered more effective when used as a con-venient means of recording text-based infor-mation via telephone and email than when it isapplied as personal e-mail voice-synthesisservice. A case of the former use might beseen, for example, in the event of rain on theoccasion of a school excursion; the teacher inthis case would send an announcement as towhether or not the excursion is postponed,notifying the system from his or her PC via e-mail on the morning of the excursion. Parentsand guardians could then call the system tohear the announcement.

6.5 Evaluation by connected organi-zations

In considering the achievements of thisproject, the evaluations of the participatingschools connected to the Experimental Net-work play an important role. With the cooper-ation of the BSNetwork Corp., a feedback sur-vey was thus conducted from February 2004to March 2004 for the five junior high schoolsnewly connected under the project, with theteachers in charge of network administrationin these schools as the respondents. The ques-tions concerned use of the network in and outof class, frequency of use, the relationshipbetween the weather and connection failures,the management of the school LAN, and othercomments. Table 1 shows the results of thesurvey.

According to the survey, the schoolsshowed a tendency to use the network more asa tool to collect information than as a means toprovide information. It also became clear thatthe schools did not use the mail server func-tion installed on the router PC but insteadrelied on web mail services. The task of sup-porting the schools in their use of the Experi-mental Network to provide information thusremains to be addressed. Use of the networkbetween the connected schools, e.g. for remoteclasses, also has yet to be established. Theseproblems should be studied in connection withapplication development, considered as thesecond phase of the project.



Results of the feedback from connected junior high schoolsTable 1

212

The frequency of use was generally highin all of the schools, which indicates that theExperimental Network can serve as an indis-pensable classroom tool. This is the source ofthe concern regarding the stability of the con-nection in many schools. One of the authors(Kanayama) currently responds to connectionmalfunctions at the user’s request, either per-sonally or with the assistance of the studentsin his laboratory. The survey neverthelessindicates that the establishment of a manage-ment system is urgently required. With theexpansion of the Experimental Network, thenumber of organizations connected to the net-work increased, and it appears the Experimen-tal Network should now move from its basicexperimental phase to an operational phase,equipped with a responsive management sys-tem. However, except for Soya Junior HighSchool, the four schools are located inWakkanai urban areas, where inexpensivecommercial connection services are available,which will gradually temper the need to relyon the Experimental Network. The cost ofensuring stable operations should certainly betaken into account when considering thefuture role of the Experimental Network.

As for the relationship between the weath-er and connection failure, although the con-nection was unstable at first, a gradual tenden-cy toward stability has been noted. As dis-cussed in 6.2.2, we know that the modifica-tion to the connection between the three hubstations (Wakkanai Hokusei, Wakkanai HighSchool, and Wakkanai Shoko High School)made in 2002 to modify the network to a trian-gular and duplex configuration was particular-ly effective. As the vulnerability of the wire-less LAN network to bad weather is no longerconsidered a problem even at Soya JuniorHigh School, where the conditions are themost severe, this issue does not now appear tomerit much concern.

Regarding the management system andother comments, we noted an unexpectednumber of complaints that the teachers couldnot operate the router PC with the FreeBSDOS. This indicates that the schools are eager

to take charge of administration of the routerPC. The early concern relating to raisingawareness among users not as passive subjectsbut as active participants in the implementa-tion of the Experimental Network thus seemsto have been addressed. In the future, it willbe necessary to consider providing FreeBSDeducation to the administrators at each schooland also to examine the possibility of con-structing the router with an OS that is easier tomanage than FreeBSD.

7 Future challenges

The equipment used in the ExperimentalNetwork will remain in use for experiments tobe performed after completion of the collabo-rative research project between the CRL andWakkanai Hokusei. This section will discussthe challenges that will remain within thefuture Experimental Network.

7.1 Challenges related to experimentsAs discussed in 6.2.2, the duplex system

has at last begun stable operation, and it isnow possible to perform experiments to deter-mine the most appropriate multicast routingalgorithm. These experiments have usuallybeen performed within stable networks andhave already provided a certain number ofconclusions. However, the findings may bedifferent for this network, which is why thisexperimental system is particularly interesting.At the same time, these experiments areimportant in the evaluation of new multicastrouting algorithms to be used in IPv6 or othernovel environments.

7.2 Challenges related to the facilityThe wireless LAN equipment now used in

the Experimental Network is mostly based onthe IEEE802.11b standard. Replacing theequipment with devices compliant with theIEEE802.11g standard should improve thenetwork specification speeds from 11 Mbps to54 Mbps without the need for antenna replace-ment.

Although the initial goals were achieved in


the first phase of the project regarding theextension of the Experimental Network, it isworthwhile to consider further extension tonearby islands in the future, as the project hasspecialized in these sorts of large, sparselypopulated areas. For example, Rishiri Island,which is in direct view of Wakkanai City, is aninteresting candidate area. Survey resultsindicate that Motorola’s 5-GHz wirelessequipment can be used to connect to RishiriIsland, though this equipment requires a userlicense and thus certain formalities will benecessary. On the other hand, there has beensome indication that ADSL services will alsobegin on Rishiri Island in 2004, in which case,it may be wise to target another site, such asRebun Island. Still, since it is unlikely thatservices such as ADSL will be provided in allareas of Wakkanai and that isolated schoolswill always remain out of reach of these serv-ices, the importance of network implementa-tion using wireless equipment—and the corre-sponding significance of the ExperimentalNetwork—will not diminish.

7.3 Relationship with other projectsWhen the current project began, inexpen-

sive Internet connection available in Wakkanaihad reached the ISDN (64 kbps) service level.This alone gave value to the fast networkthrough Wakkanai Hokusei for the schoolsinvolved. Later, development plans for a localmunicipal government intranet and for theboard of education’s School Network began toprogress. The Experimental Network thusmust be able to be positioned effectively withrespect to these municipal and educationalnetworks.

The Wakkanai City Regional IntranetInfrastructure Development Project has imple-mented a network connecting three hub facili-ties—City Hall, the Board of Education, andthe Health and Welfare Center—in addition to23 other facilities. Wakkanai City executedthe first phase of the project in 2000, with agrant provided by the former Ministry of Postsand Telecommunications, and initiated thesecond phase of the project in 2002 on its

own. By 2005, the city will conduct the thirdphase of the project, consisting of the develop-ment of the city backbone fiber-optic network;the connection of 30 public municipal facili-ties, five elementary schools, and four juniorhigh schools; and the development of 17 pub-lic access terminals. The junior high schoolsincluded in the third phase of this project arealready connected to the Experimental Net-work. Wakkanai City is aware of the exis-tence of the Experimental Network, havingreported on the activities of the ExperimentalNetwork in the context of efforts toward com-munity computerization in Wakkanai [10].However, the relationship has yet to be clari-fied between the Experimental Network andthe network to be constructed in the thirdphase of the project. Collaboration betweengrass-roots and government activities is tradi-tionally somewhat difficult. Nevertheless, itwill be a good idea to make arrangements withWakkanai City concerning the future use ofthe Experimental Network, to avoid the wasteof double or triple investment in public devel-opment. On the other hand, five years afterthe start of the project, the equipment used inthe Experimental Network has begun to com-plete its useful service life. Already severaldevices have broken down in the course of theproject, and the light sources for the laserequipment should be replaced. In consideringthe use of the Experimental Network in thefuture, the cost of updating the equipmentshould also be taken into account. Given thisbackground, we have begun discussions withWakkanai City and other parties regarding thefuture of the Experimental Network and thesharing of costs in the event of failure, forexample.

Meanwhile, all of the public high schoolsin Hokkaido are already connected to an inde-pendent public school network developed bythe Hokkaido prefecture. This network is nev-ertheless unpopular among users, and bothWakkanai High School and Wakkanai ShokoHigh School now mainly use the ExperimentalNetwork. However, the spread of other fastservices such as B Flets in Wakkanai would


214

inevitably reduce the significance of connect-ing the junior high schools to the Experimen-tal Network, as already noted. Yet the Experi-mental Network has already earned a signifi-cant local reputation, and a physical departurefrom the network will not immediately entailfull abandonment. At the same time, construc-tion of a framework that will involve a certaindegree of participation in the network centeredat Wakkanai Hokusei seems possible, even ifthe participating schools are independentlyconnected to new networks.

7.4 Application experimentsThe multicast broadcast experiment

remains as a future problem that has not evenbegun to be addressed. This experiment wasone of the targets of the project, under theassumption that the network would be used forcommunity television or for remote learningsystems. According to the plan, video cam-eras and MPEG servers were to be installed inthe connected schools, with broadcasting to beperformed via the MPEG4 multicast broadcastserver at Wakkanai Hokusei. This system wasviewed as a potential communication toolbetween the connected schools, to be used as ameans to exchange information among teach-ers as well as students. Some schools con-nected to the network have constructed a num-ber of video projects, which we believe couldbe used as broadcast content.

The impediments to this experiment todate have been discussed in 6.2.2 above, andthese problems have been resolved. On theother hand, policy and operational guidelinesfor data delivery remain as future challenges;a meeting should be organized to consider anddetermine an approach to these issues.

7.5 Stable operation of wireless LANThe feedback from the schools connected

to the Experimental Network indicates nostrong correlation between weather and theconnection stability. However, the connectionstability of 2.4-GHz band wireless LANs isknown to depend on the amount of moistureon the antenna. Wet snow will adhere to the

antenna, absorbing radio waves and reducingsensitivity. Even when the snow is not on theantenna element itself but instead adheres tothe antenna cover, some effect is inevitable. Itis windy in Wakkanai and the wind stirs up thesnow. While the temperature is generally lowin Wakkanai, and wet snow is only infrequent-ly observed, when wet snow does fall, the col-lected snow often freezes, forming a surfacefor the adhesion of further snow. Thus instal-lation of the antenna entails a thorough evalu-ation of location and disposition.

A 2.4-GHz band wireless LAN is also lim-ited in the number of available channels, andradio interference has also been recentlyobserved in the urban areas of Wakkanai. Themain causes of this problem can likely betraced to the wireless routers introduced intohomes with the spread of broadband, inter-building business communications, and wire-less services initiated through ISP. WirelessLANs that do not require a user license are notlimited in terms of number of users and thusthis trend is expected to continue. When usinga wireless LAN in backbone routing, it may benecessary to consider other frequency bandswith less interference (for example, the 5-GHzband, conforming to IEEE802.11a standards)as alternatives.

Another problem may be seen in the occa-sional unaccountable hanging of the wirelessLAN equipment currently in use. Some varia-tion in this phenomenon has been noted, butmost of the devices have been confirmed tohang. In many cases, resetting the powerallows recovery of the connection. In net-works such as the Experimental Network thatextend over a large area, relay equipment isrequired; this equipment is installed in loca-tions often difficult to access for mainte-nance—for example, in unpopulated areas orin high locations. In such cases, resetting theequipment leads to delays in recovery. It maybe necessary to consider introducing a mecha-nism to reset the power in the course of net-work traffic monitoring, or to reset powerautomatically at least once a day.


7.6 Toward advanced experimentsAs already mentioned in several sections,

the current Internet will soon have to accom-modate the introduction of IPv6. This is dueto the current global situation in which, forbetter or worse, the number of available IPaddresses has become critically low. Now theInternet faces a situation that could bedescribed as a new north-south problem.Even in the U.S., which enjoys a great numberof IPv4 addresses and where users saw littleurgency in implementing IPv6, the govern-ment network has stipulated the introductionof IPv6 within a clear time frame, with movestoward implementation expected to accelerate.Japan is relatively eager to implement IPv6,and IPv6 connections are now possible withinmany ISPs. However, IPv6 itself still offersnumerous possibilities in terms of develop-ment and the discovery of unexploited charac-teristics. As already discussed, many of therouters in the Experimental Network are PCrouters, designed to enable free experimenta-tion in connection with these new technolo-gies. We intend to introduce IPv6 into theExperimental Network in conjunction with thedevelopment of a new version of the IPv6 pro-tocol.

At the same time, a group headed by oneof the authors (Kanayama) is beginning con-nection experiments using MPLS (Multi Pro-tocol Label Switching) in collaboration withuniversities and organizations in Hokkaido(NORTH BOREO: Network Organization forResearch and Technology in Hokkaido /Broadband Orbit of Research and ExperimentOpportunity [11]). MPLS has evolved as alayer-2.5 technology in terms of logical hierar-chy, and is conceptually independent of the IP,designed based on historical lessons in theimprovement, development, or failures ofATM. MPLS is thus expected to solve theload issue in IP routing. Due to these charac-teristics, the relationship between MPLS andIPv6 also poses a significant challenge for thefuture, and the project is expected to beextending beyond the initial experimentalphase.

8 Conclusion

The environment surrounding informationand communication networks evolved signifi-cantly during the five years of the WakkanaiExperimental Community Network Project.In urban areas, ADSL and FTTH have spreadrapidly, and broadband use is now common,even in homes. The government’s introduc-tion of the e-Japan Priority Policy Programand the School Internet Project supported bythe Ministry of Education, Culture, Sports,Science and Technology and the Ministry ofInternal Affairs and Communications are twoexamples within the current movement pro-moting information and communication net-works nationwide. On the other hand, it is asdifficult as ever to use broadband services inlarge, sparsely populated areas. The costadvantages of a self-maintaining networkimplemented by wireless equipment, asdemonstrated in the current project, haveprobably not decreased. In the future, it willbe important to apply the findings of this proj-ect in the implementation of other communitynetworks.

Acknowledgements

This project was not based on CRL organi-zational policy, but instead has been promotedas part of grass-roots activities based on a bot-tom-up proposal. Yet the project could nothave continued without the understanding andsupport of CRL executives, members of thePlanning Division, and many others con-cerned, particularly at the CRL. Among thesecontributors, we received valuable assistancefrom those at the CRL Wakkanai Radio Obser-vatory, the Applied Research and StandardsDivision (responsible for the observatory), andthe Emergency Communications Group of theInformation and Network Systems Division,the host of the collaboration research team.We thank Dr. Harunobu Masuko, the execu-tive director of the Applied Research andStandards Division; Takashi Maruyama, theleader of the Ionosphere and Radio Propaga-


216

tion Group; and Hiroyuki Ohno, the formerleader of the Emergency CommunicationsGroup.

We sincerely appreciate the kind supportfor our activities in Wakkanai from ProfessorFujio Maruyama, the president of WakkanaiHokusei; Tatsuo Ueda, the dean of the Facultyof Integrated Media; Professors HiroshiSakamoto and Fumihiko Fujiki; HidenoriOkoshi from BSNetwork Corp.; Yoshinobu

Saito, the former director of Wakkanai RadioObservatory; Ichiro Yamazaki, a formerSenior Researcher; Yuko Kakizaki, a formerSE; and those concerned in the schools con-nected to the Experimental Network, at theWakkanai City Board of Education, and in thePlanning and Coordination Division, Munici-pality of Wakkanai. We would like to take thisopportunity to thank all of the above.


References1 Noriyo Kanayama, "Development and Operation of an Educational Community Network using Wireless

LAN", Proceedings of the IPSJ Autumn General Conference, 3H-3, 1998.

2 Hitomi Ohmichi et al., "Community Network using Low Power Wireless System, Proceedings of the 2001

IEICE General Conference, D-9-26, Mar. 2001.

3 Ken'ichi Sakaguchi et al (Eds.), "The Textbook to Train Teachers for Information Technology Education in

Senior High Schools", ISBN 4-501-53680-2, pp.156-157, Tokyo Denki University Press, Feb. 2004.

4 "Inter-Hokkaido High Schools Basketball Games - Relay Broadcasting on Computer Network", Hokkaido

Shimbun, Jun.23, 2000.

5 "Wakkanai Higashi Junior High School Released New Website, and Shiomigaoka Junior High School also

Has a Plan", Nikkan Souya, Mar.29, 2001.

6 "Friday Spot", NHK TV, May 18, 2002.

7 "Good Morning Report", NHK TV, May 25, 2002.

8 "Second Generation Low Power Data Communication System / Wireless LAN System", ARIB Standard

STD-T66 ver.2.0, Association of Radio Industries and Businesses of Japan, Mar. 2002.

9 "Wireless LAN System for Cold and Snowy Area -CRL Founds an Experimental Facility in Wakkanai City",

Nikkei Sangyo Shimbun, Aug.1, 2003.

10 Kazuhiro Hyugaji, "Regional Informatization Report Vol.47, Regional Informatization of Wakkanai City", Info-

Communication Journal, Vol.19, No.11, pp.26-28, Ministry of Internal Affairs and Communications, Nov.

2001.

11 Norikatsu Ooishi, "NORTH boreo MPLS -Network Organization for Research and Technology in Hokkaido /

Broadband Orbit of Research and Experiment Opportunity",

1 http://www.north.ad.jp/boreo/north-boreo-mpls_plan.pdf.

TAKIZAWA Osamu, Ph. D.

Senior Researcher, Security Advance-ment Group, Information and NetworkSystems Department

Telecommunication Technology forDisaster Prevention/Mitigation/Relief,Application Security

KANAYAMA Noriyo, Dr. Eng.

Professor, Wakkanai Hokusei GakuenCollege

Theory of Network

Progress of Wakkanai Experimental Community … over the course of five years. The report is...

Documents

Transcript of Progress of Wakkanai Experimental Community … over the course of five years. The report is...