Planning Urban Internal Transport Based on Cell Phone Data

Abstract

Based on cell phone data, mining residents’ travel characteristics and regional traffic connection characteristics has become an important technology for urban road traffic planning research. Based on cell phone data extracted from Wuxue City, this paper proposes an operability method for mining regional traffic contact intensity information based on cell phone data and an urban road network evaluation method. We analyze the travel characteristics of residents and the current situation of traffic connections between towns by constructing a planning road model of Wuxue city and find that the traffic connection of Wuxue city shows a spatial development pattern with Wuxue street as the intersection, one horizontal and one vertical, and the traffic accessibility gradually decreases from south to north. We analyze the spatial characteristics and problems of the planning of the road network in Wuxue and propose a modification plan. We verify the reasonableness of the modified plan by calculating traffic accessibility and road density. The research provides a new way to obtain spatio-temporal information about passenger flows by applying cell phone data to urban road planning.

1. Introduction

In the era of the popularization of mobile phones, scholars can extract a large number of data samples from cell phones during the process of using them for related planning research. Compared with traditional survey methods, mobile phone surveys have many advantages, such as low cost, wide coverage, and less processing time, and they have been conducted for multiple aspects of research in different fields by many scholars.

In the application research of urban transportation planning, the travel origin-destination (OD) survey is an important reference basis for solving traffic problems. It objectively and truly reflects the traffic flows between districts and the overall situation of residents’ travel. As an indispensable communication tool in people’s daily lives, mobile phones record for a long time, and at the same time, the penetration rate of mobile phones in life is high. The activities of urban residents can be monitored through mobile phones, and mobile phone data have great potential in the analysis of population travel behavior. Therefore, this paper processes the signaling data of mobile phones to extract the travel characteristics of users, thereby revealing the spatial structure of travel demand.

ArcGIS is a desktop application GIS platform that integrates spatial data display, editing, query retrieval, statistics, report generation, spatial analysis, and advanced mapping. It can comprehensively process geospatial-related information and solve complex analysis problems, and has strong information processing capabilities, which are widely used in transportation network analysis and spatial layout research of transportation facilities, which reflects the high credibility of ArcGIS spatial analysis technology analysis simulation. Therefore, this article used ArcGIS software based on cell phone data and policy documents related to urban transportation planning and urban transportation road networks. We obtain important transportation facilities data and other information from transportation planning documents, and we process cell phone data through ArcGIS to obtain the travel OD spatial distribution. We sort out the flow of people across districts in Wuxue city, count the amounts of exchanges, and get the intensity of traffic connections between districts. Then, we use this network data to construct a traffic network model of Wuxue City, analyze the problems of Wuxue City traffic planning, modify the planning deficiencies, verify the optimization of the modified plan by comparing the calculation models, such as accessibility and density analysis, and verify the rationality of the planning analysis method. Finally, we summarize the application and advantages of mobile phone data for intra-city traffic planning, and the direction of future research. The results of the OD study on residential mobility can provide help and references for local governments in urban transportation planning and resident travel planning. At the same time, we hope that our planning proposals can provide references and guidance for transportation planning in Wuxue City and other similar small- and medium-sized cities.

2. Background and Related Works

In the 1990s, scholars in different fields used road traffic volume to calculate the OD travel matrix. Fan pointed out the limitations of the optimization model represented by the maximum entropy method and proposed the Bayes statistical method [1]. Li et al. [2] proposed an iterative backstepping method based on a multi-path probability allocation model. However, these methods still have large errors. To realize low-cost and high-precision surveys of residents’ travel surveys, Shi [3] uses public bus IC card data to build a resident travel information database, infers the cardholder’s starting and ending points of public bus travel, and obtains the passengers of the bus route travel OD matrix. These methods of obtaining the travel OD matrix have made some progress, which has changed qualitative decision-making in the past. The country has begun to provide a basis for urban traffic management and planning through a quantitative analysis of various travel characteristics and needs. Li et al. used taxi OD data in Suzhou, Jiangsu Province, China to construct an index system and analysis method for taxi travel characteristics, time, and space distribution [4]. Wu et al. [5] predicted the spatial distribution of various parking needs based on vehicle ownership situations and vehicle travel OD. Lu et al. used the effective data of public transit OD trips in five typical cities of Weifang, Changzhou, Shangyu, Bengbu, and Huzhou to study the temporal and spatial distribution characteristics of residents’ public transit trips by using graphics processing and model calibration methods to provide a quantitative basis for urban public transport demand forecasting and public transportation network planning [6]. The traffic travel OD matrix is also applied to the analysis of urban public transportation, rail transit, and urban agglomeration intercity railway passenger flow [7,8,9].

With the development of communication and information technology and the popularization of mobile phones, obtaining travel OD through mobile phone data provides new ideas for transportation planning. Ma established a mobile phone traffic information collection system, using Java language to develop a J2ME-based intelligent collection of traffic travel behavior information embedded software, to realize the function of mobile phone SMS collection of travel information, and obtain various road section traffic information, dynamic travel OD matrix, etc. [10]. CDMA is a spread spectrum multi-access digital communication technology that has the advantages of strong anti-interference ability, broadband transmission, and anti-fading ability. GPSOne positioning technology is a system pushed out of the shortcomings of improving GPS positioning technology, compared with traditional GPS, which has higher accuracy, short positioning time, and can achieve indoor positioning. Huang et al. obtained the starting and ending point position information through continuous position tracking based on the CDMA communication network and GPSone cell phone location technology. In the Yangjiaping area of the Jiulongpo District, Chongqing City, they used the area judgment algorithm in the algorithm to obtain travel OD information on the basis of existing administrative divisions [11]. Based on the travel chain of cell phone location information, Wu recognizes OD points through the limitations of speed threshold, time threshold, and distance threshold [12]. In the area covered by the communication network, a specific location is usually covered by the signal overlap of several surrounding base stations. The base stations connected by mobile phones in this area will continue to change with the change in signal strength of each base station, resulting in a “ping-pong effect” phenomenon of switching between two or more base stations. In addition, to ensure the efficiency and quality of mobile phone user communication, service base stations often have a load optimization adjustment mechanism. When the load of the base station adjacent to the mobile phone is large, the mobile phone signal will be automatically switched to the base station, which is farther away and less loaded, resulting in signal drift. At the same time, the difference in the distribution density area of the base station may cause defects in the mobile phone positioning data in some areas. These problems reduce the accuracy of the phone’s data. Tang obtained the residents’ travel OD matrix to reduce the occurrence of such problems based on “space-time “ the two-level clustering stop-point recognition algorithm, and analysed the spatial distribution characteristics of OD passenger flow with ArcGIS [13]. Researchers have also summarized different methods for initial filtering and screening of positioning offset, over-range, and “noise” data in cell phone data, and combined them with the area to determine the travel chain and obtain accurate travel OD [14,15]. Mobile data processing technology continues to develop, and the accuracy of the analysis results has also improved. It is widely used to analyze the current situation of crowd travel characteristics, urban commuting, work, residence space, etc. [16,17]. Hu extracts the travel characteristics data of the current urban residents, obtains the permanent population and job position distribution, the residents’ travel rate, the total number of trips, and the OD distribution, and obtains the actual travel needs of the current urban residents [18]. Shan et al. analyzed the spatial relationship of the work-resident contact strength of permanent residents in the Jiangbei area through the Harbin cell phone data, the commute OD during the morning and evening peak hours of the working day, the current coverage of conventional buses, and the distribution of cross-river traffic, and finally proposed optimized traffic planning strategies [19]. Based on the cell phone data of Hubei Mobile users, Sha et al. established an algorithm model to analyze urban population distribution, rail transit passenger flow distribution, and regional passenger flow [20]. Foreign studies have proposed a mobile space dynamic method based on mobile network cloud travel data to estimate the starting point-midpoint journey to verify the reliability of the origin and destination travel between regions [21], and the concept of urban population flow based on the number of individuals entering and leaving a certain area per unit time. Based on the cell phone collection data, the human body trajectory grid reconstruction method is used to estimate urban crowd flow and guide traffic planning [22]. Cell phone data have the advantages of high information integration, large sample size, real-time, and high authenticity that is not affected by subjective factors and is gradually applied to future traffic demand forecasting and road planning.

Some scholars obtain the OD matrix of residents’ travel between different areas through mobile phone positioning data, then set different distance time thresholds, transfer times, and other constraints, and build a bus route planning model that calculates the bus planning route [23,24], based on the method of dividing travel transportation by the advantage of travel distance and cost. Jia used the map analysis software Transcad to obtain the bus travel matrix and forecast the public transportation demand [25]. Foreign research has developed a mobile phone location tracking algorithm from a large mobile phone network database to estimate dynamic origin-destination (od) traffic flow and travel demand data, as well as commuter traffic, and to estimate dynamic traffic demand for public transportation system planning and scheduling [26].

It can be seen from the above research status that many scholars have proposed methods for obtaining traffic travel characteristics from mobile phone data from multiple angles, but the content lacks a specific description of the methods for obtaining traffic travel characteristics (Table 1). At the same time, the form of mobile phone data acquisition is constantly changing, and new processing methods are needed to analyze regional traffic links. In addition, these studies are more specific to the urban and metropolitan levels of large urban centers, and there is a lack of research on small and medium-sized cities. In China, small- and medium-sized cities are an important force to promote the development of the national economy. This paper selects Wuxue City as the research object to provide a quantitative analysis of the development path of small and medium-sized cities in terms of transportation.

Table 1. Study comparison.

Research Focus	Comparison of Research
	Existing Research	This Article Research
Data acquisition information	Focus on describing the mining methods of cell phone data and the methods of reducing the problems of mobile phone data and strengthening the accuracy of mobile phone data [9,16,17,18,19,20,21,22].	Based on previous research, this paper focuses on the method description of how to obtain traffic flow characteristic information from mobile phone information.
Research subjects	It is mainly based on large urban areas, there is a lack of research on transportation in small and medium-sized cities [10,11,12,13,14,15,20,26].	This article focuses on most small and medium-sized cities to promote the economic development of small and medium-sized cities, thereby driving the national economic development.
Research aspect	Focus on the application of mobile phone data to study urban public transportation, and there are fewer comprehensive studies on other modes of urban transportation [23,24,25,26].	This paper not only focuses on urban public transportation, but also pays attention to urban roads, railways and water transportation, and applies cell phone data to study the planning of urban comprehensive transportation.

3. Materials and Methods

3.1. Overview of the Experimental Area

This research focuses on Wuxue city, which is located in Huanggang City, Hubei Province, at the intersection of Beijing-Kowloon Railway, Hurong Expressway, and Yangtze River. In the city, a large transportation pattern of railroads, highways, and waterways has been formed. The total area of the city is about 1200.35 square kilometers, with a registered population of 784,200, and a resident population of 644,220.

According to the population density distribution map, it can be seen that the population distribution within the municipality varies more obviously. The towns with a population density of more than 500 people/km² are mainly northern and southern towns. The urban area has the highest population density of 2442.88 persons/km², while the lowest population density is in Wanzhang Lake District and Yuchuan Town, with 351.21 persons/km² and 363.74 persons/km² respectively.

3.2. Data Sources and Processing

3.2.1. Regional Data

The following data are used for the study: 1. satellite images of Wuxue City, source spatial geographic data cloud platform; 2. administrative map data of Wuxue City with geographic information, source OpenStreetMap platform; 3. Wuxue City transportation planning documents, source government official website; 4. effective cell phone data for 20 days from 10 August 10 to 29 August 2020, collected from the whole area of Wuxue City, source operator cooperation.

3.2.2. Data Processing

First, we establish the road network model according to the following workflow (Figure 1). To carry out experimental research under the same spatial coordinate system, we first drag the city satellite image map into the ArcGIS software for coordinate projection processing, and then drag the road quantity data and mobile phone data obtained from OpenStreetMap into the ArcGIS software for the same processing. This step unifies the coordinate processing of the three with geographic information to obtain a coincident data image. Through the geographic registration tool in ArcGIS, we recombine the road planning map with the above images, establish a network database, and plot the urban planning road network, passenger stations, railway stations, ports, township points, and other data in layers according to the planning map. All roads at all levels according to the standard “Highway Route Design Specification (JTG D20-2017)” are given the speed attribute: highway driving speed is 90 km/h, first-grade road driving speed is 80 km/h, second-grade road driving speed is 60 km/h, third-grade and fourth-grade road driving speed is 30 km/h, urban arterial street driving speed is 40 km/h. The construction of the urban road network model with geographic information is shown in Figure 2. Road classification is done according to their respective tasks, functions, and traffic.

Figure 1. Planning road network model construction process.

Figure 2. Planning road network.

The total number of cell phone base stations collected in the whole area of Wuxue City are 1582, and the number of base stations was 1563 after filtering with the administrative area (Figure 3), mainly in the central city of Wuxue City (Figure 4). The sample size of the data is large but does not exclude the existence of minor deviations.

Figure 3. Base station distribution map.

Figure 4. Number of base stations in each administrative region.

3.3. Test Methods

Based on cell phone data, we mainly carried out research using ArcGIS software. We obtain the spatial distribution of travel OD, and then quantitatively analyze the main contact directions and contact strengths among cities and towns in the city. Then, we undertake an evaluation of the current planning scheme in four aspects, such as road network density and traffic accessibility, and make optimization suggestions. The research process is as follows (Figure 5).

Figure 5. Research process.

The first is spatial analysis of cell phone data. The cell phone data used in this study contain the geographical coordinates of the start and end points and the amount of traffic each day. We analyze people’s travel directions, which can be used to guide traffic planning [27]. Based on information on cell phone data and the principle of operability, this paper proposes the following analysis process to mine regional traffic contact intensity information. We first project the start and end points into the built study model. According to the direction from the starting point to the destination point, we take the average of the twenty-day flow as the daily OD flow. We connect the start and end points through the data management tool, and here is the output of all travel OD lines. We symbolically display the corresponding number of traffic, and the linear line corresponds to the travel volume from small to large and from small to large to obtain the spatial distribution of passenger flow. Then, we use the location selection tool, number the cell phone data base stations of different towns, count the travel volume from each town to other towns, sort out the traffic trip table with the town as the statistical unit, obtain the OD contact of the towns and streets in Wuxue City, and analyze the main directions of outbound travel in each township, as well as the characteristics of the city’s internal connections, for the part of the comprehensive analysis of the planned road network.

Then, we analyze the planning road network. Road density refers to the ratio of the total length of roads in a certain area to that of a country. We analyze the urban road network density using Equation (1), which is calculated as:

$$D_k=L_k/A_k$$

(1)

Among them: $D_k$ represents the road density of area k (unit: km/km²); $L_i$ represents the total length of roads in area k (unit: km); $A_i$ represents the land area of the study area k (unit: km²).

We use the ArcGIS density analysis tool to create a road density analysis map, compare the spatial distribution map of OD travel volume, and evaluate whether the traffic demand is satisfied in each area of the city. To construct an O-D cost matrix model, this paper adopts the accessibility analysis method based on the minimum resistance proposed by Allen [25]. Equation (2) presents the calculation without considering the purpose of travel and analyses the accessibility of the main activity nodes of the population in the whole city.

$$A=\frac{1}{n-1}{{\displaystyle \sum }}_{\begin{array}{c}j=1\\ j\ne i\end{array}}^n\left(d_{ij}\right),A_i=\frac{1}{n}{{\displaystyle \sum }}_{i=1}^n\left(A_i\right)$$

(2)

where $A_i$ denotes the cell phone data point accessibility, which is the average of the minimum resistance from that point to other data points, $d_{iJ}$ denotes the minimum resistance between data points i and j in terms of distance and time; A is the accessibility of the entire traffic network, which is the average of the accessibility of individual cell phone data points.

Then, the traffic station accessibility distribution map is obtained by the inverse distance weighted (IDW) interpolation method; we count the shortest path distance of each cell phone data point to reach the road intersection and calculate its average value to analyze road accessibility, which is calculated by Equation (3).

$$\mathrm{L}=\frac{1}{n}{{\displaystyle \sum }}_{i=1}^n{l}_i$$

(3)

where $l_i$ denotes the distance from data point i to the nearest road intersection.

By connecting the spatial data, we visualize the distance analysis map of each point arriving at the intersection. This paper adopts the accessibility evaluation method based on the average travel time proposed by Geertman et al. [28] to analyze the accessibility of passenger stations, railroads, and bus services, which is calculated by the following Equation (4):

$$t_i=\frac{1}{n-1}{\displaystyle \sum }_{j\ne i}{t}_{ij}$$

(4)

where $t_i$ denotes the average travel time from facility point i to departure point; $t_{ij}$ denotes the shortest travel time from facility point i to departure point j.

We calculate the minimum time required for intermodal water–land transport interchange by the shortest path calculation method and evaluate the interchange capacity of different transportation modes in the city. The synthesis of the obtained analysis results is combined with the travel OD, which points out the shortcomings of the planning scheme and suggests optimization.

Finally, the optimized road network model is constructed, and road accessibility and road density are calculated and analyzed again to obtain the range of change after the modification to verify the effectiveness of the modification.

4. Results

4.1. Analysis of the City’s Internal Transportation Links

Taking towns as units, we count the average daily travel volume of each town and the average daily travel volume across districts. Quantitative analysis was performed to obtain statistics of trips between towns (Table 2), and the map (Figure 6) of spatial linkage intensity between different regions to better determine the characteristics of regional passenger flow and provide a basis for the adjustment and construction of traffic road networks [29]. The total number of passenger trips within the city limits of Wuxue City totals 635,900•$d^{-1}$. Inter-regional travel occurs mainly in the Wuxue District and Meichuan Township.

Table 2. Statistics of trips between towns.

O/D	1	2	3	4	5	6	7	8	9	10	11	12
Dajin Town	2522	64	263	2247	358	891	2074	249	344	1138	5523	213
Wanzhanghu District	55	201	22	375	50	180	113	43	562	43	1534	41
Yuchuan Town	287	25	2236	722	133	1862	2042	131	355	252	2748	209
Shifosi Town	2221	354	719	6038	686	2992	2449	576	1117	869	11,298	495
Dafasi Town	347	44	143	708	2939	414	575	853	265	607	5528	1038
Huaqiao Town	913	174	1884	2988	406	10,072	2247	366	1694	655	8356	431
Meichuan Town	2070	125	2049	2494	613	2218	21,772	488	766	1445	9616	610
Kanjiang District	250	36	133	582	846	357	472	956	272	254	6567	764
Longping Town	343	558	350	1147	272	1648	767	261	3642	264	7009	300
Siwang Town	1155	40	251	875	598	682	1146	268	272	2159	3837	224
Wuxue District	5476	1455	2791	11,438	5570	8366	9686	6735	7034	3958	131,266	7137
Tianzhen district	235	35	197	488	1028	443	582	766	306	235	7115	3332

The first row of numbers corresponds to each township: 1. Dajin Town; 2. Wanzhanghu District; 3. Yuchuan Town; 4. Shifosi Town; 5. Dafasi Town; 6. Huaqiao Town; 7. Meichuan Town; 8. Kanjiang district; 9. Longping Town; 10. Siwang Town; 11. Wuxue district; 12. Tianzhen district.

Figure 6. Spatial linkage intensity between different regions. (a) Administrative district level. (b) Township district level.

The table fills the larger travel volume with color. The yellow markers show that the main urban area, Wuxue District, is strongly connected to Shifosi Town, Huaqiao Town, Meichuan Town, Longping Town, and Tian Town District, with the connection axis being Meichuan-Dajin-Shifosi-Wuxue, running north-south, mainly along the Mayang Expressway, and Tian Town-Wuxue-Longping at the southern end of Wuxue City, running east-west along the Yangtze River, the overall traffic connection strength shows a spatial development pattern with Wuxue Street as the intersection and a horizontal and vertical direction (Figure 7a). The green markers show that a stronger town connection is mainly between Meichuan Town, Yuchuan Town, Huaqiao Town, Shifosi Town, and Dajin Town (Figure 7b), and the traffic demand space is circular, which can increase the circular traffic facilities in order to evacuate the urban traffic gathering.

Figure 7. Spatial connection analysis. (a) City and town links. (b) Town and town links.

We count the internal travel OD passenger flow of each town (Figure 8). Meichuan town has the highest intra-town travel volume, followed by Huaqiao town and Shifosi town. For the areas with high intra-regional travel rates, considering their internal traffic demand degree is large, we can increase the public transportation operation, analyze the important traffic nodes according to the travel OD, and reasonably design the bus routes to solve intra-town traffic travel.

Figure 8. Internal travel OD statistics for each town.

4.2. Planning Road Network Density Analysis

The road network density analysis shows that the road density of each administrative district is quite different (Figure 9a), that the spatial distribution of the road network is unbalanced, and that spatial distribution has a certain concentration, mainly concentrated in the southern part of the city center area, and Meichuan town, Dajin town, and Shifosi town.

Figure 9. Analysis of road network density and travel OD passenger volume distribution. (a) Road network density (b) Administrative district level. (c) Township district level.

Comparing the planning road network density analysis with the travel OD passenger volume distribution (Figure 9a), we find that the road network density and travel OD passenger flow are imbalanced in the circled regional (Figure 9b,c). First, we need to strengthen the southeast of the towns of Wuxue City and improve the skeleton road network in the city for the town of Huaqiao, the southern part of the town of Shifosi, and other areas where the regional road network density is low. We need to strengthen road construction.

4.3. Traffic Accessibility Analysis

The accessibility of the urban traffic road network reflects the service scope of urban road networks and the convenience of urban residents’ travel, the degree of interaction between regions within the city, and to a certain extent, it can analyze the defects of urban road networks [30], so this paper studies the accessibility of traffic road networks in Wuxue City based on travel time and travel distance [31].

We obtain the accessibility distribution of the main crowd activity nodes in the city based on the analysis of cell phone data by constructing the OD matrix model (Figure 10). The crowd is mainly concentrated in the south of Wuxue City, and the accessibility of the main distribution points of crowd activity gradually decreases from south to north, and the accessibility is better in the areas with greater travel demand, but the traffic demand of Meichuan Town and Huaqiao Town, which are located in the east and north of Wuxue City is also greater from the above linkage analysis, their accessibility is poor, and the road network needs further improvement. The accessibility of the central part of the city, where highways and provincial roads are located, is high, while the eastern part of the town of Yuchuan and Huaqiao has fewer roads and poorer accessibility. The planning of the road network in the eastern and northern areas of Wuxue City needs to be further improved.

Figure 10. Accessibility analysis based on travel time.

Traffic intersections are generated by the intersection of more than two roads, which can lead to multiple directions. The average distance value of cell phone data points arriving at the nearest intersection is calculated to be 535.77 m; the standard deviation is 655.04 m. Furthermore, 67.64% of the distance is less than 500 m. The natural interruption point classification is used to obtain the road accessibility analysis map (Figure 11).

Figure 11. Accessibility analysis based on travel distance.

4.4. Accessibility Analysis of Important Transportation Facilities

Various transportation facilities in the city affect the flow of people. For small and medium-sized cities, the most important transportation facilities are passenger transport stations, railway stations, etc., and we analyze the time required for the arrival of important transportation facilities in the city through the GIS platform and evaluate its accessibility with a travel time of 5 min as a unit [32,33,34]. In constructing a service area analysis model and obtaining the accessibility distribution map of the planned passenger transport stations (Figure 12), the color in the map from red to blue shows that the time required for the arrival of important transportation facilities gradually increases. The reachable area within 5 min of passenger stations reaches 385.72 km² by calculated at the vehicle speed attribute set by each level of road, accounting for 32.1% of the city area.

Figure 12. Accessibility analysis of passenger stations.

The passenger railroad station is located in the middle of Wuxue city, and its analysis map (Figure 13) clearly shows that the accessibility of the railroad station extends outwards in a cross shape along the city highway and primary highway, and the accessibility gradually decreases, but people in the city can basically reach Wuxue station within 25 min.

Figure 13. Accessibility analysis of railroad.

The activity distribution and bus stop distribution map of the central urban area shows that the crowd activities are relatively concentrated in the south of the urban center (Figure 14a), and the bus stops are also relatively concentrated in the south. According to the 15-min and 10-min living circles defined in the 18th edition of the Urban Residential Planning and Design Standards, the accessibility of public transportation in the urban area of Wuxue was studied by analyzing the service areas of 300 m and 500 m, with the bus stop as the center. The accessibility of downtown bus stops from the center to the outside gradually decreases, but the overall accessibility of bus services is great (Figure 14b).

Figure 14. Analysis of bus station distribution. (a) Distribution of activity in the central urban area of Wuxue city. (b) Distribution of bus stops and service area.

4.5. Analysis of the Water-Land Transport Network Structure

The convenience of transfer between water and land transportation affects the passenger and freight volume of water transport, and guides the coordinated development of water-land transport to reduce the “strong land and weak water” and provide multi-modal travel for residents, which contributes to the improvement of comprehensive urban transport services and socio-economic development [34].

Wuxue City is located on the north of the Yangtze River, has good port resources, and planning also proposed the construction of Wuxue Port to further promote the development of Wuxue from a single port to the overall development of the port group to improve the city’s railroad, highway, and water transport network system. The main transport channels for Tianzhen port are the G347 national highway, the S308 provincial highway, the S411 provincial highway, and the Tianzhen railroad line. The main transport channels for Wuxue port are the G347 national highway, the Mayang highway, and the G220 national highway (Figure 15a). The number of railway stations and ports was kept separate from this for a subsequent calculation (Figure 15b).

Figure 15. Water-land transport network. (a) Railways and port planning. (b) Railway stations and port serial numbers.

Based on the accessibility evaluation of minimum impedance, we analyzed the water and land transportation network in Wuxue City. We calculated the time required for ports to reach railroad freight stations and railroad passenger stations, respectively, and then analyzed the interchangeability of the planned road network. We use the road speeds of each class to calculate the average time. The average time for each port to arrive at the freight station is 16 min and the longest time is 24 min, as shown in Table 3, and the average time taken to arrive at the Wuxue station is 18 min and the longest time is 24 min. The interchange time has improved compared with the current situation, as shown in Table 4.

Table 3. Planning port-railway freight station transfer time statistics.

Port	Railway Freight Station	Time (min)
1	1	2.21	Average: 16.04 min
1	2	24.21
2	1	3.16
2	2	23.60
3	1	13.14
3	2	20.22
4	1	15.50
4	2	20.11
5	1	15.84
5	2	22.41

Table 4. Planning port-railway passenger station transfer time statistics.

Port	Railway Passenger Station	Time (min)
1	3	23.94	Average: 18.74 min
2	3	22.24
3	3	18.56
4	3	14.15
5	3	14.48

Both the Tianzhen port area and the Wuxue port area are close to Wuxue city, water transport passenger flow is mainly concentrated in Wuxue port area. In the city’s central area, bus service coverage is mainly concentrated in the blue area. For the red area, the connection among various means of transportation is poor, resulting in inconvenient travel for residents, as shown in Figure 16. Set along the river bus lines, through water transport, traffic can quickly contact the surrounding cities.

Figure 16. Ports and city bus stop coverage distribution map.

5. Discussion

5.1. Modify the Planning Road Network

Based on the comprehensive analysis set out above, we propose the following optimization strategies.

• Strengthen effective connectivity between towns.

There are few connecting roads between towns, and there is traffic congestion. We suggest that additional road networks are needed to strengthen inter-regional links. The specific modification is shown in the figure below (Figure 17 and Table 5).

Figure 17. Add new roads.

Table 5. Modification description.

Location	Road Number	Necessity of New Road Construction
Meichuan Town	1, 2	Meet the traffic demand in Meichuan Town and strengthen its external traffic connections.
Huaqiao Town	3	Meet the traffic demand in Huaqiao Town and strengthen the connection between Huaqiao Town and Shifosi Town.
Central part of the city	4, 5	Strengthen the connection between Meichuan Town, Dajin Town, Shifosi Town and Wuxue district.

2.

Enhancing accessibility to the eastern and northern parts of the city.

The ecological zones are concentrated in the eastern and northern regions of Wuxue City, which are not conducive to the construction of new roads, but one can upgrade the existing roads, which would shorten travel time and enhance traffic accessibility. One could add town railway lines in the northeastern part of the city to strengthen the connection between the Yuchuan Township and the surrounding towns, as well as the connection between the town groups in the eastern and northern areas of Wuxue City. The red line indicates the upgrade of the original third-grade road to the second-grade road, as shown in the following design (Figure 18).

Figure 18. Modify the road grade.

3.

Improve town bus service lines.

There are some towns with a large amount of internal travel OD, such as Meichuan town and Huaqiao town. These towns can increase bus services; according to travel OD passenger flow spatial distribution, we suggest increasing bus service lines reasonably. In addition, Meichuan and Huaqiao town have larger contact with the Wuxue district, and the bus service system can be intermodal with urban environment-specific bus stops and line design, as follows (Figure 19 and Figure 20).

Figure 19. Travel OD distribution and bus route design in Meichuan Town.

Figure 20. Travel OD distribution and bus route design in Huaqiao town and Yuchuan town.

5.2. Comparative Analysis before and after the Modification of the Planning Scheme

The modified road network model is constructed to analyze road network accessibility and road density.

• Accessibility enhancement.

Based on the minimum resistance accessibility, we obtain the accessibility of the road network after the planning modification (Figure 21). We calculate the area of different reachable time ranges to compare before and after modification, and calculate the difference between before and after, and then the difference is divided by the value before modification to get the ratio of the improvement value (Table 6). In areas where passenger flow is relatively concentrated, the arrival time has been shortened.

Figure 21. Accessibility analysis based on travel time after modification.

Table 6. Comparison of accessibility indicators before and after modification.

Sort			Unit	Before	After	Promote
Accessibility coverage area	City-wide accessibility	0–15 min	km2	28.77	31.58	9.7%
		15–20 min		439.23	445.35	1.3%
		20–25 min		351.43	370.44	5.4%
		25–30 min		243.98	220.76	−9.5%
		>30 min		107.05	102.34	−4.4%
	Time required to arrive at the passenger terminal	0–5 min		330.48	339.91	2.8%
		5–10 min		420.03	467.12	11.2%
		10–15 min		95.61	93.33	−2.3%
Road intersection accessibility	Average shortest path distance		m	537.40	429.25	20.1%
	Proportion of the shortest path distance < 500		%	67.64	74.12	9.5%

The average distance value of cell phone data points arriving at the nearest intersection was calculated to be 429.24 m, and the standard deviation is 675.19 m. Additionally, 74.12% of the distance is less than 500 m. The road accessibility analysis map was obtained using a natural interruption point classification (Figure 22).

Figure 22. Accessibility analysis based on travel distance after modification.

2.

Improvement of road network density.

According to cell phone data and travel OD spatial analysis, with the suggested modifications, we reasonably increase the road in the data distribution concentration area and large travel passenger flow area. The road density analysis (Figure 23) below shows that the color of these areas has deepened, and the increase in road density enhances the decongestion capacity of the area to reduce traffic congestion.

Figure 23. Analysis of road network density after modification.

After the modification of increasing the high-grade road linking between towns and improving the internal road network, the total mileage of main roads is 1542.286 km, and the density of the main road network in Wuxue City is calculated to be 1.284 km/${\mathrm{km}}^2$, which is 109.177 km higher than the total mileage before modification, and the density of the main road network is increased by 7.3% (see in Table 7), mainly improving the road network in Meichuan Town, Huaqiao Town and Shifosi Town, the color in the passenger flow concentration area has deepened (see in Figure 24).

Table 7. Comparison of the density before and after modification.

	Before	After	Promote
Wuxue city	1.194 km/km2	1.284 km/km2	7.3%
Meichuan Town	1.090 km/km2	1.197 km/km2	9.8%
Huaqiao Town	1.114 km/km2	1.313 km/km2	17.8%
Shifosi Town	1.338 km/km2	1.581 km/km2	18.16%

Figure 24. Comparison analysis of road network density before and after modification.

6. Conclusions

The advantage of using cell phone data for road traffic planning is that mobile phone data record the trajectory of residents’ travel, which is equivalent to the passenger traffic flow between cities and towns, and this can be used to measure the degree of connection between cities and towns. This article proposes a technical framework for cell phone data to support traffic planning. The travel OD obtained from cell phone data is combined with ArcGIS analysis to analyze the main outflow direction of the city, the intensity of traffic connections between cities and towns, and the service range of important urban transportation facilities. This research provides an operable technical means for road planning and makes up for the lack of traditional data. Based on regional connection analysis, we evaluate whether the spatial distribution of planned road network density, road accessibility, and passenger bus station allocation are consistent with actual needs and use the above analysis as a guide to modify the planning road network. The rationality and effectiveness of the revised plan are verified through calculation and analysis at the end of the article. The application of cell phone data to traffic planning can objectively reflect the spatial characteristics of traffic travel and provide quantitative support for urban traffic planning.

This article proposes a technical framework for urban internal traffic applications based on cell phone data, but it still has limitations. Its analysis uses the speed of the vehicles on different grades of roads to refer to the road design specification, which leads to a lack of authenticity, and the time calculation lacks accuracy. In future work, we will further explore the application value of mobile phone data, explore how to allocate travel OD lines to the road network based on the shortest path, obtain road traffic flow, better quantitatively analyze the road network, and research methods for obtaining other information data that can be used for transportation planning.