A Congestion Control Strategy for Power Scale-Free Communication Network

Min Xiang, Qinqin Qu
2017 Applied Sciences  
The scale-free topology of power communication network leads to more data flow in less hub nodes, which can cause local congestion. Considering the differences of the nodes' delivery capacity and cache capacity, an integrated routing based on the communication service classification is proposed to reduce network congestion. In the power communication network, packets can be classified as key operational services (I-level) and affairs management services (II-level). The shortest routing, which
more » ... lects the path of the least hops, is adopted to transmit I-level packets. The load-balanced global dynamic routing, which uses the node's queue length and delivery capacity to establish the cost function and chooses the path with minimal cost, is adopted to transmit II-level packets. The simulation results show that the integrated routing has a larger critical packet generation rate and can effectively reduce congestion. The power communication network is a private communication network that meets the effective operation and management of the power system [1] . The reliability of the power communication network, which is the supporting network for the smart grid, has been the prerequisite for economical and safe operation of the smart gird [2, 3]. The power communication network is different from the public communication network because of multiple service types, which have different influences on the reliable operation of the power system. With the amount of data in the power communication network increasing explosively, the bottleneck nodes of the network are prone to congestion, resulting in that it is difficult to process the critical data in a timely manner, which endangers the safe operation of the power system [4] [5] [6] . Therefore, how to reduce the congestion in the power communication network becomes the key problem of the smart grid. Since power grid status monitoring is moving towards Internet protocols and services, the existing Internet technologies should be modified with new protocols to fit into the real-time service requirements. There are many studies about the problem of communication network congestion. The work in [7] proposed a congestion avoidance algorithm based on traffic priority for an electric power dispatching digital network, which guaranteed the QoS of high priority services. The work in [8] proposed a low risk routing method based on the fact that different types of traffic have different service importance levels to reduce the risk of the power communication network. In [9] , when the number of nodes in the power communication network increases to a certain extent, the scale-free characteristic is shown in the network. In [10], a routing algorithm based on a complex network was proposed, and the relationship between node congestion level and betweenness centrality was studied. The work in [11] proposed a protocol to work with a free-scale peer-to-peer network, by using the information about the network's architecture and by representing message dependencies at the bit level. In [12] , the node betweenness is used to compute edge weights and then routed by the shortest path. In [13] , the next-hop node was selected by studying the gravitational effect of the node in the packet transmission process, which takes the path length and the waiting time into account. In [14, 15] , a global dynamic routing strategy was proposed, which took the queue length of each node in the path as the route cost. In [16] , the optimal path was found by predicting the queues of neighbor nodes and combining the shortest path routing. In [17] , a heuristic algorithm for the optimization of transport was proposed to enhance traffic efficiency in complex networks. However, the above existing studies for the communication network do not take the scale-free characteristic of the power communication network topology and power communication service level into consideration when doing research on local congestion. The scale-free topology is characterized by a power-law distribution of the node degrees, which shows that there is a small number of highly connected nodes in the network called hub nodes. Therefore, the characteristic of scale-free networks easily causes local network congestion at hub nodes. In order to solve the local congestion problem of the power communication network, the topology model of the power communication network is established, and the delivery capacity and cache capacity of nodes are allocated according to node betweenness. Communication services are divided into two levels based on the importance of the power communication services, including key operational services of the I-level and affairs management services of the II-level. An integrated routing strategy composed of shortest path routing and global dynamic routing is proposed. Among them, the shortest path routing is adopted to transmit I-level packets, which can guarantee that the important communication packets are transmitted to the destination as soon as possible. Additionally, the global dynamic routing is adopted to transmit II-level packets to avoid the hub nodes being overloaded. Therefore, the delivery path depends on the communication services level. The minimum number of nodes between the source and destination is obtained under the shortest path routing. The cost function is set by considering the queue length and delivery capacity information of nodes in the path, and the path with the minimal cost is selected to transmit the packets under the global dynamic routing. Power Communication Network Model The high reliability power communication network architecture is composed of the core network, backbone network and access network, as shown in Figure 1 . Appl. Sci. 2017 , 7, 1054 2 of 17 dependencies at the bit level. In [12] , the node betweenness is used to compute edge weights and then routed by the shortest path. In [13] , the next-hop node was selected by studying the gravitational effect of the node in the packet transmission process, which takes the path length and the waiting time into account. In [14, 15] , a global dynamic routing strategy was proposed, which took the queue length of each node in the path as the route cost. In [16] , the optimal path was found by predicting the queues of neighbor nodes and combining the shortest path routing. In [17] , a heuristic algorithm for the optimization of transport was proposed to enhance traffic efficiency in complex networks. However, the above existing studies for the communication network do not take the scale-free characteristic of the power communication network topology and power communication service level into consideration when doing research on local congestion. The scale-free topology is characterized by a power-law distribution of the node degrees, which shows that there is a small number of highly connected nodes in the network called hub nodes. Therefore, the characteristic of scale-free networks easily causes local network congestion at hub nodes. In order to solve the local congestion problem of the power communication network, the topology model of the power communication network is established, and the delivery capacity and cache capacity of nodes are allocated according to node betweenness. Communication services are divided into two levels based on the importance of the power communication services, including key operational services of the I-level and affairs management services of the II-level. An integrated routing strategy composed of shortest path routing and global dynamic routing is proposed. Among them, the shortest path routing is adopted to transmit I-level packets, which can guarantee that the important communication packets are transmitted to the destination as soon as possible. Additionally, the global dynamic routing is adopted to transmit II-level packets to avoid the hub nodes being overloaded. Therefore, the delivery path depends on the communication services level. The minimum number of nodes between the source and destination is obtained under the shortest path routing. The cost function is set by considering the queue length and delivery capacity information of nodes in the path, and the path with the minimal cost is selected to transmit the packets under the global dynamic routing. Power Communication Network Model The high reliability power communication network architecture is composed of the core network, backbone network and access network, as shown in Figure 1 . Appl. Sci. 2017, 7, 1054 3 of 17 substations are set as access nodes. The links of provincial control center and core substations are ring connection, and the links of dispatching stations and substations are dual homing connection, as shown in Figure 2. Appl. Sci. 2017, 7, 1054 3 of 17 Figure 2. Different network architectures of the power communication network. Power communication network is a typical complex network with the scale-free characteristic. Based on the idea of graph theory, the topology model of the power communication network is built, and the modeling rules are as follows. 1. The provincial power communication network is set as a unit, and only the communication networks of 110 kV and above are considered, including power plants, substations and dispatching institutions. 2. The communication nodes and dispatch centers of the stations (power plants, substations and converter station) are abstracted as nodes without differences. 3. The communication lines between communication nodes are abstracted as edges. The communication lines are bidirectional communication, and the differences between the communication lines are ignored, that is none of the edges have weights or direction. 4. Multiple communication links in the same direction are merged, and self-loops are eliminated. Based on the modeling rules, the power communication network can be built as an undirected graph with N nodes and M edges, that is: Power communication network is a typical complex network with the scale-free characteristic. Based on the idea of graph theory, the topology model of the power communication network is built, and the modeling rules are as follows. 1. The provincial power communication network is set as a unit, and only the communication networks of 110 kV and above are considered, including power plants, substations and dispatching institutions. 2. The communication nodes and dispatch centers of the stations (power plants, substations and converter station) are abstracted as nodes without differences. 3. The communication lines between communication nodes are abstracted as edges. The communication lines are bidirectional communication, and the differences between the communication lines are ignored, that is none of the edges have weights or direction. 4. Multiple communication links in the same direction are merged, and self-loops are eliminated. Based on the modeling rules, the power communication network can be built as an undirected graph with N nodes and M edges, that is:
doi:10.3390/app7101054 fatcat:2stjidsvcbdifirmp35sinoa2y