GPS log analyses for circular tour activity

GPS log analyses for circular tour activity Ustijana Shikoska 1,1, Dancho Davchev 2, Risto Rechkoski 3, Elena Petrovska 4, Jordan Sikoski 5 1 EVN Macedonia, KEC Ohrid, Makedonski Prosvetiteli 11, 6000 Ohrid, R. Macedonia ustijana@t-home.mk 2 University Professor at University "Sts. Cyril and Methodius"- Skopje, Faculty of Electrical Engineering and Information Technologies, Head of Computer Science Department, Head of Open and Distance Learning Center, Karpos II, 1000 Skopje, R. Macedonia, http://odlskopje.etf.ukim.edu.mk, etfdav@feit.ukim.edu.mk 3 University Professor at Sv. Kliment Ohridski Univeristy-Bitola, Faculty of Tourism and Hospitality-Ohrid, Kej Marshal Tito 95, 6000 Ohrid, R. Macedonia, reckoice@t-home.mk 4 Tourist Information Center Municipality of Bitola, R. Macedonia, petrovska.e@gmail.com 5 Ministry of Education and Science, R. Macedonia, jordans@t-home.mk Abstract. An analyzing method of circular tour activity based on GPS in order to realize suitable tourism promotion for current tourism is given in our paper. This paper focuses on circular tour activity with rent-a-car. Our method concerns basic activity information which includes the information about destinations and movements on circular tour, based on GPS log. The analysis for circular tour activity is performed considering the basis of the acquired basic activity information. Tourism attractiveness level of the town of Ohrid and major tour route for circular tour route are analyzed using stochastic model. The effectiveness of proposed method is confirmed by the experiment using GPS log collected from actual tourists performing circular tour in Ohrid, Republic of Macedonia. Also, circular tour activity is analyzed and comments about the results are given in our paper. Keywords: Global Positioning System (GPS), GPS log, position localization, activity information, stochastic model, GIS (Geographic Information System) 1 Introduction Global Positioning System (GPS) together with Geographic Information System (GIS) have found their own great usage also in analyzing circular tour activities in the latest growth industry tourism. Tourism has been expected as a growth industry and national project for tourism promotion to survey the actual conditions has been performed in R. Macedonia. It is necessary to construct tourism strategy and policy considering surveyed information. Statistical information such as inflow number of tourists on each transportation and the operation rate of hotels has been mainly surveyed few years ago [4] [6]. So far, the actual tourism condition could be grasped from such statistical information because the group tour with journey course is fixed in advance with major tour Macedonian style. The number of group tours decreased and the number of personal

circular tours increased rapidly. It is difficult to grasp the actual condition of tourism from conventional statistical information because each tourist goes sightseeing on the basis of individual tour plan. Therefore, the detailed information about circular tour activity on each tourist has to be surveyed in order to implement suitable tourism promotion for current tourism environment. There is conventional survey, in which questionnaires are given to tourists where they describe the activity information on circular tour to the questionnaires on their basis of their remembrance. But, the number of tourists is extremely large in this kind of survey, also there are some problems with a raise of the omission of the description, so it s difficult to describe accurate time information. The cost of converting questionnaire to electronic data is high. Considering previously said, there was a need to develop the method in which the detailed and accurate circular tour activity information can be easily obtained and analyzed. In our paper, an analyzing method of circular tour activity is proposed, which is based on GPS log, in order to realize effective tourism promotion, suitable for current tourism environment. In our research, circular tour activity using a rent-a-car is considered because tourists which use rent-a-car rapidly increase with the increase of personal circular tour. In this method, firstly, the sequential position data has been collected, on circular tour using rent-a-car, by using Global Positioning system (GPS). The basic activity information on circular tour is presumed from this sequential position data called GPS log. The basic activity information includes the information about the destinations and movements on circular tour. Recent GPS can accurately detect present position in real time. However, in the position detection using the standard GPS, error arises under the effect ionosphere, atmospheric distortion, multipath and condition of GPS receiver. GPS receiver can not detect the position when it can not receive the radio signals sent by GPS satellites. [1][2]. Our method is robust against various errors and situations and it can presume accurate basic activity information from GPS log. In this method, the analyses of tourism attractiveness level on each city and major tourism route are performed by using the accurate basic activity information and stochastic model. Effective tourism promotion could be implemented by making it possible to easily acquire and analyze the information about circular tour activity from GPS log only. The effectiveness is confirmed through the experiments using GPS log collected from actual tourists in Ohrid. Circular tour activity analyses and results discussions are given also. 2 GPS Log Collection In this paper there is a tendency to collect sequentially position data on circular tour by using GPS and then effectively analyze circular tour activity on the collecting data. Therefore, we have developed GPS log collection device in order to collect sequential position data. Figure 1 shows a GPS log collection device in this study.

Fig. 1. GPS log collection device This device is developed on the basis of GPS receiver produced by Garmin Ltd. The size of the device is 17cm (Length) x 23cm (Width) x 120cm (Height). The device operates by dry cells, it can continuously run seven days. It can record 10,000 log data of the position information into its memory. Log data can be easily transferred to the personal computer through RS232C interface. Time for recording log data could be set at one second intervals. 3 Circular Tour activity Analyses Method 3.1 Basic activity information GPS log consists date, time, latitude and longitude. GPS log is collected at an arbitrary time interval on circular tour. This method presumes basic activity information on circular tour from the collected GPS log. This information is about destinations and movements on circular tour as basic activity information. Arrival time, departure time and position for each destination; the distance and time for each movement among destinations are also presumed. The information about city (town) transition is presumed. GPS can accurately detect present position in real-time. Considering position detection using standard GPS, error arises under the effect of ionosphere, atmospheric distortion, multipath and condition of GPS receiver. GPS receiver couldn t detect the position when it couldn t receive radio signals sent by GPS satellites. These errors and situations cause the noise in GPS log. The proposed method is robust against various errors situations and can presume accurate basic activity information from GPS log. 3.2 Calculation of primitive information among GPS log The time interval, distance and speed between a GPS log on time t and the GPS log on time t+1 are calculated. The time interval is calculated by using the information of date and time in GPS log. The distance among GPS log is calculated by Hubeny s distance formula. The speed among GPS log is calculated on the basis of time interval and distance. A city (town) in which GPS log on time t was recorded is estimated. In this estimation, the distance between GPS log and each city hall in the survey is calculated. GPS log was recorded in the city (town) having the minimum distance between the position in the GPS log and the city hall [6] [7].

3.3 Error log filtering GPS log having extreme error is judged by using two criteria based on speed and distance information. The procedure judges that a GPS log on time t has extreme error if the speed between GPS log on time t and GPS log on time t-1 is more than 80km/h. GPS log on time t has extreme error if the speed is higher that 60km/h and the distance is more than maximum length of expressway in the survey area between GPS log on time t-1 and the GPS log on time t. If it is judged that a GPS log on time t has extreme error, the position is modified to the average position of GPS log before and after it. 3.4 Extraction of destinations and movement information This procedure presumes and extracts the regions in a sequence of GPS log which are corresponding to the situations that tourist stays at destination or that tourist moves to destination. In this procedure the presumption of the regions which are corresponding to the situations for stay is performed and the rest of regions in the sequence is treated as the situation for movement. A GPS log on time t is temporarily presumed as the situations of stay when the speed in the GPS log is less than speed parameter Sp l. GPS log on time t is temporarily presumed as the situations of stay, if the speed of a GPS log on time t is less than the speed parameter Sp 2 and if the speeds of GPS log before and after it are less than a speed parameter Sp l. These criteria are designed in order to adapt to normal errors and unexpected large errors included in GPS log. When the time interval in the region of GPS log which are temporarily presumed as the situation of stay is more than a time interval parameter lp, the region is formally extracted as the region corresponding as the situation of stay. The rest of regions of GPS log which are not presumed as the situation of stay are extracted as the regions corresponding with the movement situation. In the extraction of destination and movement information, the speed among the GPS log is utilized as the information for their presumption. Using the speed information to presume destination and movement, the information of them can be extracted if the GPS receiver can not record GPS log at the preset time interval because the speed in GPS log become low. The proposed method adapts to various situations of GPS receiver existence inside the buildings, tunnels (in the Ohrid town center area there are three tunnels), etc. 3.5 Adjustment of wrong extraction The noise in each GPS log is considered in the extraction of destination and movement information. There is the possibility that wrong extractions are caused by sequential large noises. As such wrong extractions, it is likely that a region for a stay is divided into some wrong regions for a stay by the errors which are similar to the situations of movement or by continuous large errors [3] [5]. To solve such wrong extraction we introduce the adjustment procedure for wrong extraction. In this adjustment, firstly, the average latitudes and the average longitudes, the average positions, in the regions which are corresponding with destinations and movements are calculated. Next, the time intervals among destinations are calculated. If the

distance among the average positions on destinations is less than time parameter Ct, destinations are unified to one destination. If the distances between average positions of two destinations and the average position of movement which is put between them are less than distance parameter Cd, the regions of movement and two destinations are treated as the region for one destination. The wrong extraction may occur where there are many high-rises because amount of error in the position detection is large on such place. In this procedure, the consideration of the urban area with many high-rises is performed. The vicinity of the city hall on each city is treated as urban area because the city hall of each city (town) generally exists in the central part of city where there re many high-rises. The distance between the average position of destinations and the positions of city halls is calculated. If the shortest distance is less than distance parameter Ud, it s judged that the destination is in urban area. If all target destinations in this adjustment exist in urban area, twice the preset values are used in the parameter Cd and Ct. The robustness of proposed method is promoted by the adjustment of wrong extraction and the parameter tuning using area information. 3.6 Analyzing the circular tour activity In this method, the analyses of circular tour activity is performed by using the presumed basic activity information. Two analyses are performed; tourism attractiveness level analyses and major tour route analyses. The tourism attractiveness level analyses can make clear numerically the quality for tourism which each city (town) has. Tendency of movement on circular tour can be grasped by the analyses of major tour route. Effective tourism promotion can be implemented by using these analyses. Many tourists visit the city having high attractiveness level for tourism, and then they stay there for long time. Tourists stay often in the city because the city has many tourist attractions. In the proposed method, the tourism attractiveness level on each city is calculated on the basis of the expected length and frequency of stay. Major tour route is estimated o the basis of transition probability among cities. The calculation method of the tourism attractiveness level on each city and major tour route are described as follows next. Calculation of a city transition probability matrix A city transition probability matrix P is calculated on the basis of the city transition information in basic activity information. The major tour route is estimated by using this probability matrix. The elements in this matrix are calculated as follows: p l f ( i, j) =, f ( i, j) = Mk( i, ) n (1) k= 1 f ( i, j) i, j j j= 0

Here, p ij represents the transitions probability between city l and city j on tourist k. l indicates the number of tourists. Calculation of a matrix for stay length on each city and a matrix for stay frequency on each city A matrix S for the length of stay on each city and the matrix R for the frequency of stay on each city are calculated by using the basic activity information. The elements of each matrix are calculated by the following formula: s i, j l Ik( i, j), l k = 1 k = 1 = l Tk( i, j) l (2) Ik(l,j) and Tk(l,j) denote the total length of stay and the total frequency of stay on tourist k after moving from city l to city j. Calculation of the matrix for expected stay length on each city and a matrix for expected stay frequency on each city A matrix IE for expected stay length on each city and the matrix TE for expected stay frequency on each city are calculated by using a city transition probability matrix P, a matrix S for stay length and a matrix R for stay frequency. The elements of these matrixes are calculated as follows: ie ij pij sij, teij = pij rij (3) Calculation of attractiveness level on each city In an expected stay length matrix IE and an expected stay frequency matrix TE, total expected stay length is i on city l and total expected stay frequency ts i are calculated, respectively. The tourism attractive level A i on city l is calculated on their basis. n n is i = ieij, tsi = teij, Ai = isi j= 0 j= 0 ts i (4)

4 Analyses of Circular Tour Activity from Actual Tour Log Basic activity information accuracy verification. The basic activity information on circular tour is presumed and then the analyses for circular tour is performed on the basis of this information, in our method. Firstly, the accuracy of the basic activity information presumed by the proposed method is verified. This verification is performed using GPS log which were collected from actual tourists in Ohrid, Republic of Macedonia. The invitation of subjects was performed in the Ohrid airport poplar branch, rent-a-car, from April, 2009 to September, 2009. The GPS log collection device was installed to a rent-a-car which the subjects rent and they perform tourism without any limitations. The GPS log was recorded at intervals of a minute. A questionnaire was given to the subjects and they were asked to describe the destination name, arrival time, the time interval of stay, the purpose of stay, when rent-a-car was stopped. In this invitation, the GPS logs of 125 subjects were collected. The basic activity information was extracted from the collected GPS logs by using this method. The parameters were: Sp1 = 3,9km/h, Sp2 = 7,3km/h, lp = 300sec, Cd = 127,5m, Ct = 178sec, Ud = 4km. These parameters were determined considering accuracy measurement experiment result of GPS log collection device and the main issue is position localization based on standard GPS. Destinations information was compared to the method on the questionnaire. The coincidence of destinations was judged on the basis of the arrival time, the time interval, destination name, destination position. In this method, the regions of GPS log which were not detected as destinations are considered as regions for the movement. Only destinations were compared to them on questionnaire because movement information is correct if destination information is correct as well. 5 Results The comparison results are given in Table 1, as follows. Table 1. Comparison results of stay information Coincidence rate Extraction impossible Un-described destinations W.ex. 89.5 % 102 4.1 0,4 Coincidence rate indicates the percentage of coincidence between the destinations extracted by the proposed method and them described on the questionnaire. Extraction impossible represents the number of destinations which could not be extracted by this method, although they were described in the questionnaire. Undescribed destinations indicate the number of the destinations which were extracted

by the proposed method although they were not described on the questionnaire by subject. These destinations were examined considering GPS log data on the map position localization. Wrong extraction was also examined considering GPS log data and the position on the map. In the comparison result, the value of coincidence rate is high, the values of extraction impossible and wrong extraction were low. It s obvious that our method could extract the accurate destination information in circular tour. The effective survey of circular tour activity using questionnaire is difficult. The adequate analyses results can be obtained considering the extracted basic activity information. The analyses of circular tour activity was performed considering acquired basic activity information. The tourism attractiveness level was calculated twice. Once, all basic activity information is utilized for analyses. Second time, the considered information as stay for lodging is removed. Table 2 indicates the results of tourism attractiveness level analyses. Table 2. Results of tourism attractiveness level analyses Lodging City name Attractive Expected Expected level stay Length stay frequency (min.) (times) Ohrid 1642.2 768.6 2.4 Bitola 500.7 307.9 1.4 Struga 116.5 105.4 1.0 Oteshevo 42.1 55.6 0.4 No lodging City name Attractive Expected Expected level stay Length stay frequency (min.) (times) Ohrid 268.7 112.2 2.0 Bitola 145.1 115.4 1.3 Struga 32.1 31.6 1.0 Oteshevo 10.0 30.1 0.1 In Table 2, results for city of Ohrid and four more cities (towns) in R. Macedonia which have high attractiveness level are shown. The expected length of stay and the expected frequency of stay are indicated here. The city of Ohrid and the city of Bitola have indicated high value because the invitation of subjects was performed in this region. The value of attractiveness level for Ohrid decreased when the stay for lodging was not utilized for analyses. Obviously, many tourists stay in Ohrid for lodging and Ohrid is made as base of tourism. Struga and Oteshevo indicated high value with exception of neighboring city. Figure 2 shows the result of major tour route analyses.

Fig. 2. Major tour route analyses result The transitions among cities are drawn. The transition probability value is more than 0.5. An effective tourism promotion can be performed if the cooperation for tourism is implemented among the cities connected with high transition probability. 6 Conclusion In this paper, an analyzing method had been proposed of circular tour activity, based on GPS log in order to produce effective tourism promotion. The effectiveness is confirmed through the experiment using the GPS log collected from actual tourists in town of Ohrid, R. Macedonia. Considering experimental results, the accurate circular tour activity information is obtained, only with GPS log, by using the proposed method, which confirms great GPS usage in many fields of interest. The tourism attractiveness level and major tour route in Ohrid were discussed, by performing analyses usage in extract information, which is of big importance for our town. In the proposed method, latitude and longitude information is provided as the destination position localization information. It can be implemented Geographic Information System (GIS) technology to the proposed method. But, detailed information, like destination name should be given for this implementation. As a future work, the effectiveness of proposed method can be done by estimating the contents of tourism from GPS log.

References 1. Tanabe J., Lee Y. H.: Characteristics of positioning data for monitoring behavior, 7 th World Conference of ITS, Torino (2007) 2. John e. Harmon, Anderson S.: The Design and Implementation of Geographic Informational Systems 3. Mohinder S., Lavrence R.: Global Positioning Systems, Internal Integration and Navigation 4. Reckoski R.: A new effort for tourism growth in Western Balkan Region: National strategy for tourism development in the Republic of Macedonia 2008-2012, 26 th EUROCHRIE Conference, Dubai (2008) UAE 5. A. Pereira-Neto,C. Unsihuay, and O.R. Saavedra, Efficient evolutionary strategy optimisation procedure to solve the nonconvex economic dispatch problem with generator constraints, IET Gener.Transm.Distrib., 1, (2), pp. 366 368, (2007). 6. Pawlak, Z. and Skowron, A. Rough sets: Some extensions Information Sciences, An International Journal. 177(1), (2007), 28-40. 7. Ustijana Shikoska., Dancho Davchev, Jordan S.: Sensitivity analyses for GPS in land-vehicle navigation, IEEE International Conference on Wireless and Mobile Computing, Networking and Communication, WiMob 2008, Avignon, France, October (2008). Acknowledgements I want to express my gratitude to my honored Professor and Mentor Dancho Davchev, PhD, Full Professor at Faculty of Electrical Sciences and Computer Engineering, Sv. Kiril I Metodi - University in Skopje, R. Macedonia, Head of Computer Science Department, Head of Open and Distance Learning Center and Chairman of SoftCOM 2009, for his professional guidance, useful advices and help for GPS knowledge and implementation experience in general, also in this work. My special gratitude to honored Professor Gordana Petrovska Reckoska, PhD, Full Professor at Faculty for Tourism and Hospitality in Ohrid, Sv. Kliment Ohridski University in Bitola, R. Macedonia, Department Chair, my loving Mother, for professional knowledge, useful advices and great help in general, especially in this work. Special thanks to Roleks rent-a-car in Ohrid, Tourist Agencies in Ohrid and Tourist Information Center Municipality in Bitola for the professional cooperation in GPS collection and research for our work.