阅读说明：本技术 一种面向高动态卫星网络的虚拟网络映射方法 (Virtual network mapping method for high dynamic satellite network ) 是由 刘江 周宇柯 黄韬 何晓春 于 2019-12-10 设计创作，主要内容包括：本发明公开一种面向高动态卫星网络的虚拟网络映射方法,包括以下步骤：建立卫星底层网络架构模型；通过地面控制器以及承载在卫星上的控制器采集网络拓扑信息,并且维护一张虚网统计拓扑和一张记录底层网络资源状态的抽象统计拓扑；当一个虚网请求达到时,控制器如果接受该虚网请求,则根据所使用的映射算法给虚网分配底层资源,并且更新抽象拓扑和虚网统计拓扑；当一个虚网服务完成时,则从底层节点释放已分配的资源。该算法由地面站和高轨道卫星承载主要的网络控制功能,统一对虚拟网络请求进行处理并且映射到物理网络,在出现网络故障的时候主动感知链路状态变化传输给控制器,控制器提取断开链路上的虚网请求,进行重映射。(The invention discloses a virtual network mapping method for a high dynamic satellite network, which comprises the following steps: establishing a satellite bottom layer network architecture model; acquiring network topology information through a ground controller and a controller loaded on a satellite, and maintaining a virtual network statistical topology and an abstract statistical topology for recording the resource state of a bottom layer network; when a virtual network request arrives, if the controller receives the virtual network request, allocating bottom layer resources to the virtual network according to the used mapping algorithm, and updating the abstract topology and the virtual network statistical topology; when a virtual network service is completed, the allocated resources are released from the underlying node. The algorithm carries main network control functions by a ground station and a high orbit satellite, uniformly processes virtual network requests and maps the virtual network requests to a physical network, actively senses the state change of a link and transmits the state change to a controller when a network fault occurs, and the controller extracts virtual network requests on a disconnected link and remaps the virtual network requests.)

1. A virtual network mapping method facing a high dynamic satellite network is characterized by comprising the following steps:

establishing a satellite bottom layer network architecture model, and establishing a virtual network, namely an architecture model of a virtual network;

acquiring network topology information of a bottom network and a virtual network through a ground controller and a controller loaded on a satellite, and maintaining a virtual network statistical topology and an abstract statistical topology for recording resource states of the bottom network;

when a virtual network request arrives, if the controller receives the virtual network request, allocating bottom layer resources to the virtual network according to the used mapping algorithm, and updating the abstract topology and the virtual network statistical topology; when a virtual network service is completed, the allocated resources are released from the underlying node.

2. The method for mapping a virtual network to a high dynamic satellite network according to claim 1, wherein the step three includes allocating bottom layer resources to the virtual network according to a mapping algorithm used, and the specific process includes: firstly, detecting whether topology change occurs, if so, extracting the affected requests and adding the affected requests into a request queue, recovering bottom-layer resources occupied by the requests, and if not, directly implementing the following operations on the request queue; then mapping virtual network requests in sequence, wherein the mapping of the virtual network requests comprises node mapping and link mapping, virtual network node mapping is carried out based on a greedy algorithm, and link mapping is carried out based on a K shortest path algorithm;

the specific mapping steps of the virtual network request are divided into two steps, firstly, a greedy algorithm is used for virtual network node mapping, and all nodes in the current virtual network request are sequenced from large to small according to a CPU (Central processing Unit) before the node mapping; after each virtual node in the virtual network is mapped successfully, mapping the link according to a K shortest path algorithm to find K shortest paths between two nodes in a bottom layer network, if one path meets the bandwidth requirement of the virtual link, mapping the path to the bottom layer path, and if the K paths do not meet the bandwidth requirement, failing to map the link; if the mapping fails, the request is put into a waiting queue, and if the failure times exceed a preset remapping threshold parameter, the request is directly rejected.

3. The method according to claim 1, wherein the satellite underlying network architecture model is:

the bottom layer of the whole satellite is a weighted directed graph G^S＝(N^S，E^S)，N^SRepresenting a collection of underlying nodes, E^SRepresents a set of underlying links; for N^SArbitrary node n in the set^SHas a symbol c (n) representing the size of its node CPU resource^S) For E^SEach link e in the set^S(i, j) each has a symbol b (e) representing the size of its link bandwidth^S) And a symbol l (e) with a side link delay size^S) Wherein i and j respectively represent a source node and a destination node of the link, and the path set of all the underlying networks is P^SIs represented by P^S(s, t) represents the set of all paths from the source node s to the destination node t;

the architecture model of the virtual network is as follows:

virtual network representation as a directed graph G^V＝(N^V，E^V)，N^VRepresenting a collection of underlying nodes, E^VRepresents a set of underlying links; for N^VArbitrary node n in the set^VHas a symbol c (n) representing the size of its node CPU resource^V) For E^VEach link e in the set^V(i, j) each has a symbol b (e) representing the size of its link bandwidth^V) (ii) a The virtual network request is denoted as V_i＝(G_V，t_s，t_e)，t_sTime, t, representing the start of the virtual network_eA time indicating the end of the virtual network; the ground controller distributes bottom layer physical resources for the virtual network when the virtual network starts, and the ground controller recovers the bottom layer physical resources distributed to the virtual network when the virtual network ends.

4. The method for mapping a virtual network to a high-dynamic satellite network according to claim 2, wherein in step three, the mapping of the virtual network requests in sequence is ordered from large to small according to request priority.

5. The method for mapping the virtual network facing the high-dynamic satellite network according to claim 2, wherein in the third step, a greedy algorithm is adopted for virtual network node mapping, if the current virtual network request does not specify access to a satellite node, a node with the largest remaining resources is selected from the satellite nodes which can be accessed as a first node for mapping, the remaining virtual nodes in the request are selected from a set meeting the requirement of a virtual node CPU, and the node with the largest remaining resources is mapped.

6. The method for mapping virtual network of high dynamic satellite network as claimed in claim 5, wherein for satellite nodes without specified access, a resource parameter R of an access satellite is introduced_A(n_s) The parameter is used for measuring the resource residual situation when the satellite is used as an access satellite, and is defined as follows:

R_N(n^S) Representing the remaining CPU resource of the current node, T (n)_s) Representing the number of remaining antennas of the current node,representing the remaining total bandwidth, R, of the link to which the current node is connected_A(n^S) The larger the value of the virtual network mapping information is, the more resources related to the virtual network mapping are possessed by the node, the success rate of virtual network request mapping can be improved, and the utilization rate of bottom layer resources is improved;

a mapping process of a virtual network request, denoted by the symbol M, is to be performed on a virtual network G^VMapping to underlying physical network G^SMeanwhile, the node resources and link resources required by the virtual network should be smaller than the rest of the resources of the underlying network;

M(G^V)：(G^V，c(N^V)，b(e^V))→(G^S，R_N(n^S)，R_E(P))。

Technical Field

The invention relates to the technical field of aerospace communication, in particular to a virtual network mapping method for a high dynamic satellite network.

Background

With the rapid development of the aerospace field, the cost of launching satellites becomes lower and lower. Meanwhile, the satellite mobile communication system has the characteristics of wide coverage area, small influence by landform and landform, no influence by natural disasters and the like, and is very suitable for application of offshore defense, overseas communication, emergency communication in major disasters and the like. Therefore, each country starts to deploy its own satellite network system, the number of satellites starts to rise sharply, the satellite system starts to carry complex and variable communication services, such as space communication, data relay, meteorological observation, deep space communication, relay transmission and the like, and future satellite systems have more perfect network functions. However, since satellite-borne resources of satellites are limited, node resources and bandwidth resources on the satellite are very precious, and it is necessary to virtualize network resources and computing and storing capabilities on the satellite, improve the utilization of the satellite-borne resources, and implement customized network services for spatially complex tasks.

In 2005, a team of the professor Larry Peterson of princeton university proposed a concept of network virtualization, which integrates and abstracts infrastructure resources of a network into a plurality of virtual networks without reversing the existing internet architecture, and provides a programmable interface to simultaneously map a plurality of virtual networks with different network topologies, which are isolated from each other, on a common physical network. Network virtualization is implemented by virtual network mapping, which refers to mapping virtual network requests onto a physical network efficiently and meeting the requirements of the virtual network on various physical resources (such as node computing power, link bandwidth, etc.). Virtual network mapping needs to solve various problems such as resource constraint, admission control, online request, topology diversity, and the like. Efficient virtual network mapping can improve the utilization rate of physical network resources, so that more virtual networks can be operated on the same basic network, development of a novel network technology is facilitated, and reduction of operation cost of an operator is facilitated.

The virtual network mapping mode can be divided into online processing and offline processing. The off-line processing algorithm knows the physical resources required by the virtual network before mapping and then maps all virtual networks to the underlying network at once. In practical applications, the details of the virtual network cannot be predicted, so the algorithm has no significance for practical applications. The latter algorithms are mostly online processing algorithms, which generally process randomly arriving virtual network requests with different required resources and update the underlying resource state in real time according to the arrival and departure of the virtual network. The algorithm for searching the same composition proposed in the literature is also an online processing algorithm, and the algorithm has high efficiency under the condition that the arrival scale of the virtual network is small, but under the condition that the arrival scale of the virtual network is large, the time complexity of the algorithm is greatly improved due to excessive backspacing caused by search failure.

Because the satellite network has fast dynamic change and the network topology changes all the time, an online processing algorithm is adopted in the satellite network to collect network topology data in real time to calculate virtual network mapping. The technology adopts a satellite network architecture fusing a software defined network. The SDN is a mode for realizing network virtualization, and the core design concept of the SDN is to realize the separation of a control plane and a data forwarding plane of a network, and a centralized controller is used for controlling underlying network hardware facilities, so that the flexible scheduling and control of network flow are realized. Therefore, the SDN technology can effectively reduce the device load, assist the network operator to better control the infrastructure, and reduce the overall operation cost, which becomes one of the most promising network technologies.

In a satellite network system, an SDN controller is mainly deployed on the ground, and collects inter-satellite topology information and network resource use conditions of a satellite network by using a ground station, and then performs centralized calculation to allocate network resources to each virtual network request. The satellite nodes are also responsible for realizing part of control functions, wherein the satellite positioned in the high orbit can be controlled by the high orbit satellite under the condition that a ground controller cannot directly control certain satellites due to the large coverage area. Whenever the network topology of the satellite changes, the controller will extract the virtual network requests affected by the topology change, recalculate the mapping scheme for these requests, and allocate resources.

At present, researches on virtual network mapping algorithms are few, but most of the researches are carried out on scenes of an opposite network, how to improve the utilization rate of underlying network resources and the success rate of virtual network mapping is considered, and mapping algorithms specially designed for high-dynamic scenes similar to satellite networks are few.

The distributed star swarm network virtual resource arbitration mechanism research provides a distributed star swarm network virtual resource arbitration mechanism, aiming at the problem of virtual resource conflict in distributed star swarm network virtual resource management, a game theory and a path splitting method are introduced, and conflict resources are recalculated and distributed when virtual network resources conflict. When a virtual network resource conflict occurs in a certain path, firstly, a path splitting method is adopted to calculate the idle resources of the parallel link of the path, and then the conflicted resources are distributed to the parallel path. If the residual idle resources of the parallel path cannot meet the resources of the conflict request, the idea of the game method is utilized, and the priority of the request is combined, so that the bandwidth is reasonably distributed to the virtual network with the conflict, and the reasonable use of the distributed star swarm network resources is ensured.

The mechanism runs on a distributed star swarm network virtual resource management framework, and the framework adopts a layered and domain-divided mechanism. The global scheduling center is responsible for maintaining a global resource information table and a topology information table, and the mapping information is updated after the mapping of one request is completed. After the minimum logic topology change period of each mobile satellite network is finished, trying to adjust the bandwidth resources requested by the virtual network with lower priority according to the bandwidth use condition. The domain-divided scheduling center refers to a management node in each mobile satellite region and is responsible for collecting the resource use condition of the satellites in the domain and maintaining four tables of resource information, topology information, virtual network information and mapping virtual network in the domain. The satellite node maintains a virtual link resource table, periodically senses the use condition of surrounding resources to update, and reports the resource information of the surrounding links of the node to the domain-divided scheduling center in real time.

The disadvantages of the above technique are:

(1) cannot adapt to actual scene of satellite network

The resource arbitration mechanism is optimized only for three possible conflict scenarios in virtual network mapping, namely mapping conflict caused by change of bottom layer resources in the mapping process of the global scheduling center, mapping conflict caused by lack of consideration of resources except for a CPU and a bandwidth, and conflict caused by change of network request resource demand.

(2) High dynamic property incapable of adapting to satellite

Since most satellites of the satellite network run in a low orbit, the movement speed is high, the topology of the network changes very frequently, and when a link is disconnected, affected virtual networks need to be extracted and then the network requests are remapped. When a link is established, the link state table needs to be updated in time to achieve efficient utilization of the link. The virtual resource scheduling mechanism does not consider the request remapping when the bottom link topology changes, which can cause the virtual network request mapping failure.

(3) Large amount of calculation on the satellite

The general satellite node needs to collect the resource usage of the surrounding links at regular time and report to the satellite of the domain-divided scheduling center. In addition, when the local global scheduling center fails to calculate and allocate virtual resources, the local global scheduling center needs to take over to complete the operation, however, the path splitting method adopted by the mechanism needs to calculate multiple paths by using the OSPF algorithm, and the complexity of the algorithm is very high. These functions all require good on-board computing power from the satellite. However, communication satellites are mainly used for forwarding, the on-satellite resources are very limited, and if a large amount of resources are used for calculation and network condition sensing, the whole satellite network is congested.

(4) Large resource consumption and low reliability

The distributed star swarm network virtual resource arbitration mechanism adopts a layered scheduling mechanism which is divided into a global scheduling center on the ground and a domain scheduling center on the star. When the satellite is used as a local scheduling center, a plurality of spare satellites are needed to maintain high reliability of the system. The whole satellite network has a plurality of domain-divided scheduling centers, and the cost is high. In addition, since the satellite network covers the whole world, but the coverage of the ground station is limited, a plurality of ground stations need to be deployed around the world to realize true global management and control.

Disclosure of Invention

Aiming at the characteristics that a satellite communication network has rapid dynamic change, the bandwidth of links between satellites is not equal in two directions, all satellites cannot be completely covered by a ground station and the like, the virtual network mapping mechanism for the high dynamic satellite network is designed, the ground station and the high orbit satellite carry main network control functions, and virtual network requests are processed and mapped to a physical network in a unified manner. The controller actively detects to obtain the topology of the satellite network, meanwhile, the satellite nodes can also play an auxiliary role, the state change of the link is actively sensed and transmitted to the controller when the network fails, and the controller extracts the virtual network request on the disconnected link to remap.

In order to solve the technical problem, the invention provides a virtual network mapping method for a high dynamic satellite network, which comprises the following steps:

establishing a satellite bottom layer network architecture model, and establishing a virtual network, namely an architecture model of a virtual network;

acquiring network topology information of a bottom network and a virtual network through a ground controller and a controller loaded on a satellite, and maintaining a virtual network statistical topology and an abstract statistical topology for recording resource states of the bottom network;

when a virtual network request arrives, if the controller receives the virtual network request, allocating bottom layer resources to the virtual network according to the used mapping algorithm, and updating the abstract topology and the virtual network statistical topology; when a virtual network service is completed, the allocated resources are released from the underlying node.

Further, in the third step, the bottom layer resource is allocated to the virtual network according to the used mapping algorithm, and the specific process is as follows: firstly, detecting whether topology change occurs, if so, extracting the affected requests and adding the affected requests into a request queue, recovering bottom-layer resources occupied by the requests, and if not, directly implementing the following operations on the request queue; then mapping virtual network requests in sequence, wherein the mapping of the virtual network requests comprises node mapping and link mapping, virtual network node mapping is carried out based on a greedy algorithm, and link mapping is carried out based on a K shortest path algorithm;

the specific mapping steps of the virtual network request are divided into two steps, firstly, a greedy algorithm is used for virtual network node mapping, and all nodes in the current virtual network request are sequenced from large to small according to a CPU (Central processing Unit) before the node mapping; after each virtual node in the virtual network is mapped successfully, mapping the link according to a K shortest path algorithm to find K shortest paths between two nodes in a bottom layer network, if one path meets the bandwidth requirement of the virtual link, mapping the path to the bottom layer path, and if the K paths do not meet the bandwidth requirement, failing to map the link; if the mapping fails, the request is put into a waiting queue, and if the failure times exceed a preset remapping threshold parameter, the request is directly rejected.

Further, the satellite underlying network architecture model is as follows:

the bottom layer of the whole satellite is a weighted directed graph G^S＝(N^S，E^S)，N^SRepresenting a collection of underlying nodes, E^SRepresents a set of underlying links; for N^SArbitrary node n in the set^SHas a symbol c (n) representing the size of its node CPU resource^S) For E^SEach link e in the set^S(i, j) each has a symbol b (e) representing the size of its link bandwidth^S) And a symbol 1 (e) with a side link delay size^S) Wherein i and j respectively represent a source node and a destination node of the link, and the path set of all the underlying networks is P^SIs represented by P^S(s, t) represents the set of all paths from the source node s to the destination node t;

the architecture model of the virtual network is as follows:

virtual network representation as a directed graph G^V＝(N^V，E^V)，N^VRepresenting a collection of underlying nodes, E^VRepresents a set of underlying links; for N^VArbitrary node n in the set^VHas a symbol c (n) representing the size of its node CPU resource^V) For E^VEach link e in the set^V(i, j) each has a symbol b (e) representing the size of its link bandwidth^V) (ii) a The virtual network request is denoted as V_i＝(G_V，t_s，t₆)，t_sTime, t, representing the start of the virtual network₆A time indicating the end of the virtual network; the ground controller distributes bottom layer physical resources for the virtual network when the virtual network starts, and the ground controller recovers the bottom layer physical resources distributed to the virtual network when the virtual network ends.

Further, in step three, the mapping of the virtual network requests in sequence means that the requests are sorted from large to small according to the priority of the requests.

Furthermore, in the third step, performing virtual network node mapping by using a greedy algorithm means that if the current virtual network request does not specify access to a satellite node, a node with the largest remaining resources is selected from the satellite nodes which can be accessed by the request as a first node for mapping, and the nodes with the largest remaining resources are selected from a set which meets the requirements of a virtual node CPU for mapping.

Furthermore, for satellite nodes without specified access, a resource parameter R of an access satellite is introduced_A(n_s) The parameter is used for measuring the resource residual situation when the satellite is used as an access satellite, and is defined as follows:

R_N(n^S) Representing the remaining CPU resource of the current node, T (n)_s) Representing the number of remaining antennas of the current node, ∑_es_eL(n^S)b(e^S) Representing the remaining total bandwidth, R, of the link to which the current node is connected_A(n^S) The larger the value of the virtual network mapping information is, the more resources related to the virtual network mapping are possessed by the node, the success rate of virtual network request mapping can be improved, and the utilization rate of bottom layer resources is improved;

a mapping process of a virtual network request, denoted by the symbol M, is to be performed on a virtual network G^VMapping to underlying physical network G^SMeanwhile, the node resources and link resources required by the virtual network should be smaller than the rest of the resources of the underlying network;

M(G^V)：(G^V，c(N^V)，b(e^V))→(G^S，R_N(n^S)，R_E(P))。

compared with the prior art, the invention has the following implementation effects:

(1) more applicable to satellite networks

Aiming at a real application scene of a satellite network, a measurement mode of accessing residual resources of a satellite is introduced by considering that covered satellite nodes need to be selected when a satellite user accesses, so that the satellite nodes accessed by virtual network users have resources such as a large CPU, available link bandwidth and antenna number, and the probability of successful mapping is improved.

Secondly, aiming at the characteristic that the satellite communication network adopts simplex communication, the condition that bidirectional bandwidths of bottom links are not equal is considered during mapping, and the mapping method is closer to a real mapping scene.

And thirdly, because the topology of the satellite network periodically changes, the topology changes are sensed in advance, the affected requests are extracted, and mapping is carried out again, so that the time of user service interruption is shortened, and the user service quality is improved.

(2) Low delay, load balancing

When the link mapping is carried out, the residual bandwidth and the time delay of the link are comprehensively considered, so that the mapping result has the advantages of low link time delay and network load balance, the utilization rate of bottom layer resources of the whole network is improved, the user experience is improved, and the virtual network request acceptance rate is improved.

Drawings

Fig. 1 is a schematic structural diagram of a satellite network virtualization model according to the present invention.

Detailed Description

The invention will be further explained with reference to the drawings and the specific examples below: