MPE Mathematical Problems in Engineering 1563-5147 1024-123X Hindawi Publishing Corporation 694956 10.1155/2013/694956 694956 Research Article The Analysis of the Properties of Bus Network Topology in Beijing Basing on Complex Networks Zhang Hui 1 Zhao Peng 1 Gao Jian 1 Yao Xiang-ming 1, 2 Lin Yi-Kuei 1 School of Traffic and Transportation Beijing Jiaotong University Beijing 100044 China njtu.edu.cn 2 State Key Laboratory of Rail Traffic Control and Safety Beijing Jiaotong University Beijing 100044 China njtu.edu.cn 2013 26 3 2012 2013 01 01 2013 27 02 2013 2013 Copyright © 2013 Hui Zhang et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

The transport network structure plays a crucial role in transport dynamics. To better understand the property of the bus network in big city and reasonably configure the bus lines and transfers, this paper seeks to take the bus network of Beijing as an example and mainly use space L and space P to analyze the network topology properties. The approach is applied to all the bus lines in Beijing which includes 722 lines and 5421 bus station. In the first phase of the approach, space L is used. The results show that the bus network of Beijing is a scale-free network and the degree of more than 99 percent of nodes is lower than 10. The results also show that the network is an assortative network with 46 communities. In a second phase, space P is used to analyze the property of transfer. The results show that the average transfer time of Beijing bus network which is 1.88 and 99.8 percent of arbitrary two pair nodes is reachable within 4 transfers.

1. Introduction

Complex networks have been successfully used in many real complex systems since the researches of small-world networks and scale-free networks [1, 2]. Many real complex systems have been well studied which include the actor network , WWW [2, 4], protein networks , and power grid [1, 3]. During the last few years, transport networks, such as subway network , airport network , and street network , have been studied by the complex approach.

As an important part of urban transport systems and a trip mode to alleviate the traffic congestion, bus network has been studied by an increasingly large number of researchers. Sienkiewicz and Holyst studied the public transport in 22 Polish cities and found that the degree distribution of these network topologies followed a power law or an exponential function . Xu et al. analyzed the three major cities of China, and they found that there is a linear behavior between strength and degree . Soh et al. contribute a complex weighted network analysis of travel routes on the Singapore rail and bus transportation systems .

In this paper, we investigate the Beijing bus network (BBN) with 722 lines and 5421 nodes. The data can be achieved from the Internet (http://www.mapbar.com/search/). In the network, the nodes stand for bus stations and edges (links) are the bus line connecting them along the route. In order to analyze the static properties of BBN, we use the so-called space L and space P to represent the BBN. The space L is mainly used to analyze the properties of degree distribution, cluster, the average shortest path, degree correlation, and community structure. The space P is mainly used to analyze the average transfer time.

This paper is organized as follows. In Section 2, we give the description of the construction of the network. Section 3 analyzes the main properties of the BBN with space L. In Section 4, we analyze the average transfer time of the BBN and use space P to calculate the smallest transfer times between any nodes in the network. Discussion and conclusions are given in Section 5.

2. The Construction of the Network

For easy calculation, there are some assumptions followed.

The station name is the unique identification in the network. Do not account for the condition that some stations have identical names but different parking place.

The stations have slight difference between upstream line and downstream line. In this paper, the construction of the undirected network is based on the stations of the upstream lines.

This paper does not consider the number of links between two stations and the frequency of bus; that is, it does not consider the weight of the network.

The bus network is usually represented by space L, space P , space B , space C . This paper will study and the Beijing bus network based on space L and space P. Space L consists of nodes which stand for the bus stations and a link between two nodes exists if they are consecutive stops on the route. Space P is a link formed between any two nodes of a line. Figure 1 gives a schematic representation of space L and space P.

Illustration of space L (a) and space P (b).

3. The Main Properties of the Beijing Bus Network under Space L

Here we represent network as a graph G=(V,E), where V is the set of nodes and E is the set of edges (links). G is described by the N×N adjacency matrix {eij}. If there exists an edge between nodes i andj, eij=1; otherwise eij=0. N is the number of nodes in the network. Figure 2 gives the Beijing bus network topology graph with 5421 nodes and 16986 links.

The graph of Beijing bus network topology.

3.1. Degree

The degree ki of node i is defined as the number of nodes that connected with the node i, which reflects the importance of the node i. In this paper, it refers to the number of bus stations with direct bus connecting with the current bus station. After calculation, the largest degree of the BBN is 21, the smallest is 1, and the average degree of all nodes is 3.13 which means one station averagely connects 3-4 stations in Beijing bus network. Table 1 shows partly the bus stations of larger degree of BBN.

The stations of Beijing bus network with larger degree and the value of degree.

Serial number Bus station degree
1 Sanyuanqiao 21
2 Liu Li Qiao Dong 20
3 Beijing Xi station 19
4 Liu Li Qiao Bei Li 18
5 Beijing Zhangdong 16
7 Xi Dao Kou 16
8 Chong Wen Men Xi 15
9 Guang An Men Nei 14
10 Zuo Jia Zhuang 14
11 Qianmen 14
12 Dabei Yao Nan 14
13 Tianqiao 13
14 Ma Dian Qiao Nan 13
15 Bei Tai Ping Qiao Xi 13
16 Si Hui station 13
17 Deshengmen 13
18 Xi Bei Wang 13
19 Xiyuan 13
20 Xin Fa Di Qiao Bei 13

Here, we studied the proportion of the stations with the degree from 1 to 21. Figure 3 shows the degree distribution of the stations in the BBN, and we found that it follows a shifted power law distributionf(k)=1.914*Γ(k)/Γ(k+2.47), k>2, k is the degree, and the number of stations with degree 1 accounts for 4.48%. It also shows that the degree of 99% of all station is smaller than 10.

The degree distribution of Beijing bus network.

3.2. The Average Shortest Path

The average shortest path is the property to reflect the efficiency of information circulating on the network. It is defined as  l=2*i>jdij/(N*(N-1)),  l is the average minimum shortest number of steps between all pairs of nodes, and dij is the shortest path between node  i and nodej.

3.3. Cluster Coefficient

Cluster coefficient is an important property of characterizing the local cohesiveness of the current node or the extent to which the nodes in the network are clustered together. In the BBN, clustering coefficient reflects the ease of the bus transport among the neighboring bus stations of the current one. It is defined as  Ci=Ai/(ki(ki-1)/2), where Ci is the cluster coefficient of nodei, Aiis the actual number of links between the neighbor nodes of the current node, and ki is the degree of node i. The cluster coefficient of the network isC=iGCi/N.

3.4. Efficiency

The efficiency is the property to characterize the capacity of traffic, and it can be calculated with the formula  E=2*i>j(1/dij)/(N*(N-1)), where dijis the shortest path between nodeiand nodej .

3.5. Degree Correlation

Degree correlation reflects the relationship between the degrees of nodes. Nodes with high degree tending to be connected with nodes with high degree are called assortativity. In contrast, nodes with high degree which have the tendency to be connected with low degree are called disassortativity. It can be calculated with the formula  (1)r=M-1ijiki-[M-1i(1/2)(ji+ki)]2M-1i(1/2)(ji2+ki2)-[M-1i(1/2)(ji+ki)]2, whereris correlation coefficient and jiandkiare the degrees of the nodes at the ends of theith links, with i=1,,M.

3.6. The Community Structure

It is usually found that there are many communities in one complex network; within the community, there are many links, but between the communities, there are fewer links . Newman and Girvan  gave a measure Qcalled modularity. For a division with g communities, then define a g×gmatrix e whose component eijis the fraction of edges in the original network that connects nodes in community ito those in communityj. The modularity is defined to beQ=ieii-ijkeijeki=Tre-e2, where xindicates the sum of all elements of x. It can be achieved that0Q1. Q=0 indicates the community structure is not stronger than it would be expected by random chance. The larger the modularity is, the stronger the community structure is.

Table 2 shows the main properties of BBN. We can see that the average shortest path is 20.03, which means people can reach destination by averagely taking 20 stops, and it is obvious that the BBN exhibits a small-world property. The cluster coefficient of BBN is 0.142, which means the BBN is a sparse network. Furthermore, the correlated coefficient is 0.185, which validates the result in the paper [m6] that when the number of nodes is larger than 500, the network is usually assortative. In addition, the community structure of BBN is so obvious and the modularity is 0.905 with 46 communities.

The value of the main properties of Beijing bus network.

Network parameters Value
Number of nodes (N) 5421
Number of lines (L) 722
Average degree (Ak) 3.13
Average shortest path (l) 20.03
Cluster coefficient (C) 0.142
Efficiency (E) 0.066
Correlation coefficient (r) 0.185
Number of communities (Z) 46
Modularity (Q) 0.905

4. The Transfer Property of Beijing Bus Network under Space P

The transfer capacity is an important index to evaluate the performance of a bus network, and travelers always expect that they can reach the destination through the least number of transfers. In this paper, the average minimum transfer time is used to evaluate the performance of the transfer capacity. Usually, travelers cannot reach the destination without transfer for a long distance trip, and the minimum transfer time between any two nodes is specific. The average minimum transfer time is the average among all pair nodes.

G = ( L , V ) is used to represent the specific bus network, whereL is the line,Vis the bus station, and Gis a L*N matrix{bij},bij=1if lineistops at stationj; otherwisebij=0.

Table 3 shows the specific network given in Figure 1. We can achieve the minimum transfer time using Table 3. For example, (1) search  V6V7. Because there is no directed line between the node V6 andV6, it needs transfer. From Table 3, we can see that b16=b13=1 through  L1, and b23=b27=1 through  L2, so we can achieveV6L1V3L2V7; that is, travelers need transfer at V3from L1toL2. (2)SearchV10V13. We can get the path V10L2V3L1V5L3V13through two transfers.

The specific network of Figure 1.

V 1 V 2 V 3 V 4 V 5 V 6 V 7 V 8 V 9 V 10 V 11 V 12 V 13
L 1 1 1 1 1 1 1 0 0 0 0 0 0 0
L 2 0 0 1 0 0 0 1 1 1 1 0 0 0
L 3 0 0 0 0 1 0 0 0 0 0 1 1 1

Using the aforementioned method, we can get the minimum transfer time between any two nodes and calculate the average minimum transfer time. But it becomes very hard when the scale of network is becoming huge. In this paper, we use the space P to solve the problem. Firstly, we need to construct the network under space P, where the weight of the network is 1. Secondly, the Floyd algorithm is used to achieve the shortest path between any two nodes. The shortest path value is the minimum line number that needs to use and the transfer time is the needed line number minus 1. Figure 4 gives the illustration of the aforementioned example. Figure 4(a) shows V6can reach V7by using the two dotted lines. Figure 4(b) shows V6can reach V7by using the three dotted lines.

The illustration of transfer network under space P.

Here, we study the BBN. Table 4 shows the stations that have most lines. It is found that the station that owns most lines is San yuan qiao which has 47 lines.

The stations that have most line and the line numbers.

Serial number Station Line numbers
1 Sanyuanqiao 47
2 Liu Li Qiao Bei Li 40
3 Beijing Xi Zhan 39
4 Zuo Jia Zhuang 38
5 Liu Li Qiao Dong 34
6 Liu Li Qiao Nan 33
7 Dongzhimen Wai 31
8 Gong Zhu Fen Nan 31
9 Bei Da Di 30
10 Bei Tai Ping Qiao Xi 29
11 Jing An Zhuang 29
12 Xiajia Hutong 29
13 Qing He 29
14 Xiyuan 29
15 Beijing Zhangdong 28
16 Xi Bei He 28
17 Xiju 28
18 Si Hui Zhan 28
19 Liangmaqiao 28
20 Mu xi yuan qiao dong 27
21 Yan Huang yishu Guan 26
22 Yuquanying Qiao Xi 26
23 Dongwu Yuan 25
24 Guang An Men Nei 25
25 Qianmen 25
26 Wanshou si 25
27 Mu Xi Yuan Qiao Xi 25
28 Kandan Qiao 25

Figure 5 shows the distribution of the proportion of the stations and the amount of line number; the result shows that it follow the exponent distributionf(x)=0.79*e-0.67x, where xis the number of lines. It is found that most stations of BBN own less than 10 lines and only 9 stations own more than 30 lines.

The distribution of a station’s amount of lines.

In this paper, the transfer time of BBN is studied by using space P. From Table 5, we can see that the most pair nodes are reachable through one or two transfers, and 99.85 percent of the pair nodes is reachable within four transfers and the average minimum transfer timeatr=1.88. Usually, the larger the atr is, the worse the performance of the bus network is. In general, atrcannot be more than 2; otherwise, we can consider that the performance of the bus network is bad and travelers’ trip is inconvenient. The transfer time of BBN is a little large and there is a room for improvement.

The proportion of the transfer time of Beijing bus network.

atr 0 1 2 3 4 4 more unreachable
Proportion 0.0168 0.283 0.553 0.141 0.0047 0.00013 0.00137
5. Conclusion

In this paper, space L and space P are used to analyze the static properties of Beijing bus network. Space L is used to research the main topology properties of the Beijing bus network. The results show the Beijing bus network has small cluster coefficient, scale-free feature, and assortative correlation and the community structure is obvious. Moreover, we research the transfer property using space P. The result shows that the accessibility of the Beijing bus network is good and the average minimum transfer time is 1.88, which is a little large. A convenient bus network needs less transfers and high performance, and how to reduce transfer time and enhance the bus network dynamical performance is a valuable research.

Acknowledgment

This paper is supported by the Specialized Research Fund for the Doctoral Program of Higher Education of China (20120009110016).

Watts D. J. Strogatz S. H. Collective dynamics of “small-world” networks Nature 1998 393 440 442 Barabási A.-L. Albert R. Emergence of scaling in random networks Science 1999 286 5439 509 512 10.1126/science.286.5439.509 MR2091634 Ravasz E. Barabási A. L. Hierarchical organization in complex networks Physical Review E 2003 67 7 026112 10.1103/PhysRevE.67.026112 Adamic L. A. Huverman B. A. Power-law distribution of the World Wide Web Science 2000 287 5461 2115 10.1126/science.287.5461.2115a Maslov S. Sneppen K. Specificity and stability in topology of protein networks Science 2002 296 5569 910 913 2-s2.0-0037012880 10.1126/science.1065103 Latora V. Marchiori M. Is the Boston subway a small-world network? Physica A 2002 314 1–4 109 113 2-s2.0-0036940215 10.1016/S0378-4371(02)01089-0 Barrat A. Barthélemy M. Pastor-Satorras R. Vespignani A. The architecture of complex weighted networks Proceedings of the National Academy of Sciences of the United States of America 2004 101 11 3747 3752 2-s2.0-1642308591 10.1073/pnas.0400087101 Guimerà R. Mossa S. Turtschi A. Amaral L. A. N. The worldwide air transportation network: anomalous centrality, community structure, and cities' global roles Proceedings of the National Academy of Sciences of the United States of America 2005 102 22 7794 7799 10.1073/pnas.0407994102 MR2167709 ZBL1135.90309 Jia T. Jiang B. Building and analyzing the US airport network based on en-route location information Physica A 2012 391 15 4031 4042 Jiang B. A topological pattern of urban street networks: universality and peculiarity Physica A 2007 384 2 647 655 2-s2.0-34548052471 10.1016/j.physa.2007.05.064 Sienkiewicz J. Holyst J. A. Statistical analysis of 22 public transport networks in Poland Physical Review E 2005 72 11 046127 10.1103/PhysRevE.72.046127 Xu X. P. Hu J. H. Liu F. Liu L. S. Scaling and correlations in three bus-transport networks of China Physica A 2007 374 1 441 448 2-s2.0-33751228132 10.1016/j.physa.2006.06.021 Soh H. Lim S. Zhang T. Fu X. Lee G. K. K. Hung T. G. G. Di P. Prakasam S. Wong L. Weighted complex network analysis of travel routes on the Singapore public transportation system Physica A 2010 389 24 5852 5863 2-s2.0-77958492678 10.1016/j.physa.2010.08.015 Seaton K. A. Hackett L. M. Stations, trains and small-world networks Physica A 2004 339 3-4 635 644 10.1016/j.physa.2004.03.019 MR2091390 von Ferber C. Holovatch T. Holovatch Y. Palchykov V. Public transport networks: empirical analysis and modeling The European Physical Journal B 2009 68 2 261 275 2-s2.0-64149094322 10.1140/epjb/e2009-00090-x Latora V. Marchiori M. Efficient behavior of small-world networks Physical Review Letters 2001 87 4 198701 2-s2.0-35348957264 198701 Newman M. E. J. Assortative mixing in networks Physical Review E 2002 89 4 208701 10.1103/PhysRevLett.89.208701 Newman M. E. J. Detecting community structure in networks The European Physical Journal B 2004 38 2 321 330 2-s2.0-2942564706 10.1140/epjb/e2004-00124-y Newman M. E. J. Girvan M. Finding and evaluating community structure in networks Physical Review E 2004 69 15 026113 10.1103/PhysRevE.69.026113