CN114499644A - Load balancing routing method based on accurate link state feedback - Google Patents

Load balancing routing method based on accurate link state feedback Download PDF

Info

Publication number
CN114499644A
CN114499644A CN202210127411.1A CN202210127411A CN114499644A CN 114499644 A CN114499644 A CN 114499644A CN 202210127411 A CN202210127411 A CN 202210127411A CN 114499644 A CN114499644 A CN 114499644A
Authority
CN
China
Prior art keywords
link
node
time
delay
congestion
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202210127411.1A
Other languages
Chinese (zh)
Inventor
杨玉璠
张冰
张奭
齐晓鑫
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Xidian University
Original Assignee
Xidian University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Xidian University filed Critical Xidian University
Priority to CN202210127411.1A priority Critical patent/CN114499644A/en
Publication of CN114499644A publication Critical patent/CN114499644A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B7/00—Radio transmission systems, i.e. using radiation field
    • H04B7/14—Relay systems
    • H04B7/15—Active relay systems
    • H04B7/185—Space-based or airborne stations; Stations for satellite systems
    • H04B7/18578—Satellite systems for providing broadband data service to individual earth stations
    • H04B7/18584—Arrangements for data networking, i.e. for data packet routing, for congestion control
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B7/00—Radio transmission systems, i.e. using radiation field
    • H04B7/14—Relay systems
    • H04B7/15—Active relay systems
    • H04B7/185—Space-based or airborne stations; Stations for satellite systems
    • H04B7/1851—Systems using a satellite or space-based relay
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W28/00—Network traffic management; Network resource management
    • H04W28/02—Traffic management, e.g. flow control or congestion control
    • H04W28/0289—Congestion control
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W28/00—Network traffic management; Network resource management
    • H04W28/02—Traffic management, e.g. flow control or congestion control
    • H04W28/08—Load balancing or load distribution
    • H04W28/09—Management thereof
    • H04W28/0925—Management thereof using policies
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W40/00—Communication routing or communication path finding
    • H04W40/02—Communication route or path selection, e.g. power-based or shortest path routing
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W40/00—Communication routing or communication path finding
    • H04W40/24—Connectivity information management, e.g. connectivity discovery or connectivity update
    • H04W40/248—Connectivity information update

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Physics & Mathematics (AREA)
  • Astronomy & Astrophysics (AREA)
  • Aviation & Aerospace Engineering (AREA)
  • General Physics & Mathematics (AREA)
  • Computing Systems (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)

Abstract

The invention discloses a load balancing routing method based on accurate link state feedback, which mainly solves the problems that the existing method has inaccurate control on congestion conditions, low notification efficiency and overlong end-to-end time delay of delay sensitive flow caused by neglecting different service requirements in the flow, and the realization scheme is as follows: 1) dividing a time slice and obtaining a group of static topological graphs; 2) performing routing planning on each time slice; 3) monitoring the abnormal condition of a link in real time in the process of planning the route: if the link is congested, different congestion processing is carried out according to the service requirement, so that the link is not congested any more. The invention reduces unnecessary signaling overhead by processing the congested link; the detour of a part of packets which do not pass through the congestion link is avoided; the delay of delay sensitive traffic is reduced, and the method can be used for load balancing in a low-orbit satellite network.

Description

Load balancing routing method based on accurate link state feedback
Technical Field
The invention belongs to the technical field of communication, and further relates to a load balancing routing method which can be used for realizing load balancing routing selection in a low earth orbit satellite network.
Background
In satellite networks, routing problems have always been the focus of research by experts and scholars both at home and abroad in order to make efficient and rational use of expensive satellite network resources. With the development of satellite network technology, the network scene is gradually diversified, and the traffic demand in the satellite network is rapidly increased. Due to the influence of geographic factors and production activities, the density distribution of satellite network users and traffic is not uniform, so that heavy traffic is accumulated on a specific area or some data links, and network resources in other areas are idle, thereby causing congestion and causing rapid deterioration of network performance. Load balancing is an effective method for balancing workload and extending the life cycle of a network in a network, and aims to properly allocate network resources so that the network resources can better provide network services. In general, effective load balancing techniques can improve system performance and prevent congestion or overload, so load balancing also becomes a non-negligible problem in satellite network routing.
Load balancing routes of a satellite network can be divided into global information-based load balancing routes and local information-based load balancing routes. The global load balancing collects the global load state information of the network, and makes a flow balancing decision in the whole network range according to the current network state. The local load balancing allows each satellite to independently make a load balancing decision according to the local load state information of the satellite, so that the load balancing in a local range is realized.
In the existing load balancing routing algorithm, the routing scheme adopting global load balancing comprises a satellite link state routing algorithm SLSR, a selective link routing algorithm ALR, a compact explicit multi-path routing algorithm CEMR and a selective iteration Dijkstra algorithm SIDA; the routing scheme adopting the local load balancing comprises an explicit load balancing algorithm ELB, a signal lamp-based intelligent routing algorithm TLR and a hybrid global-local load balancing routing algorithm HGL. Wherein:
the SLSR algorithm only distributes uncertain network information such as queuing delay and link state by calculating the propagation delay of the snapshot off-line, thereby avoiding that the flow is concentrated to the shortest path, but the congestion regulation reaction is slow.
The ALR and CEMR algorithms both use a multi-path routing strategy to distribute traffic over multiple selectable paths, so that traffic is distributed more uniformly, but traffic distribution is random, which may cause congestion in a split path.
The SIDA algorithm reduces the repeated use frequency of the nodes by changing the traversal sequence of the Dijkstra algorithm, and inhibits the number of newly congested links after shunting by using a selective switch load balancing strategy SSLB, but the time delay may be increased.
The ELB algorithm avoids sending packets to a heavy-load node by exchanging congestion information between adjacent nodes, can quickly relieve local congestion, but can cause unnecessary detours of some packets due to inaccurate judgment of a congestion link, thereby increasing time delay. Although the ELB algorithm can perform differentiated processing on different service types of traffic, the processing of delay-sensitive traffic is too simple, and only by means of the shortest path, a situation that packet delay is too large due to too large queuing delay may occur.
The TLR algorithm can enable each packet to obtain an approximately optimal transmission path by introducing a traffic light mechanism to dynamically adjust a route, and can avoid congestion, but because a plurality of queues are introduced and the judgment method is complex, the TLR algorithm has higher requirements on the computing power and the storage space of nodes.
The HGL algorithm controls congestion by global planning and local real-time adjustment of routing flow, and a global and local mixed management mode is adopted to enable network performance to be better, but a global processing strategy needs to use estimation of network flow change conditions, and only the flow change in the next short time is considered when the HGL algorithm carries out flow prediction, so that the HGL algorithm is not accurate as a flow prediction basis for long-time flow management.
Disclosure of Invention
The present invention aims to provide a routing method based on accurate link state feedback to adjust the transmission path of a packet according to the link congestion situation, reduce the traffic sent to the link to be congested, reduce the queuing delay and the number of packet losses generated thereby, and ensure the adaptability to the emergency and more service scenarios by processing different types of services.
The technical idea of the invention is as follows: by accurately monitoring the congestion state of the link in real time and feeding the congestion situation back to the neighbor node, the neighbor node bypasses the packet passing through the congested link, thereby avoiding further aggravating the congestion and more effectively realizing the network load balance. For different service types, service requirements of different services are met as much as possible by carrying out differentiated services.
According to the above thought, the technical scheme of the invention comprises the following steps:
(1) dividing the satellite network into equally spaced time slices in time to turn the high-speed dynamic masking of satellite movement into a series of static topologies;
(2) and (3) carrying out route planning on the time slices divided in the step (1):
(2a) taking propagation delay and queuing delay as real-time link cost measurement, and adopting Dijkstra algorithm to calculate the shortest path of each time slice;
(2b) updating the routing table according to the shortest path, namely writing the next hop corresponding to the shortest path into the next hop field of the routing table to obtain a new routing table;
(2c) monitoring the link state in real time: if the link congestion occurs, executing the step (3), otherwise, continuously monitoring the link state;
(3) and (3) shunting the condition that the link is congested:
(3a) setting up queues at four output links of each satellite node, calculating real-time occupancy rates mu (t) of the queues, and dividing congestion states of the satellite output links into three states of idle, busy and busy according to the occupancy conditions of the queues;
(3b) when a certain output link is changed from idle to busy, the satellite searches a routing table of the satellite to find all routing table items of which the next hop passes through the output link, and sends a destination node set corresponding to the table items to all other neighbor nodes except the neighbor node corresponding to the output link of the node through a warning packet; after receiving the warning packet, the neighbor node searches for a substitute route which does not comprise a congestion link for the packet to the node in the destination node set, and updates the substitute route into a route table;
(3c) when a certain output link is changed from a busy state to a busy state, the satellite sends a congestion notification packet to a neighbor node which has sent a warning packet, so that the corresponding neighbor node sends the flow to the congestion link according to the set flow dividing ratio x, and transmits the residual flow with the ratio of 1-x through the found alternative route;
(3d) when a certain output link is changed into an idle state from a busy state, after the shunting of the step (3b) continues to be continued for a period of time tau, the routing table is restored to the state of the routing table obtained in the last step (2 b);
(3e) if different types of service requirements exist in the network, dividing the flow in the network into delay sensitive flow and non-delay sensitive flow, bypassing the non-delay sensitive flow, if the link is still in a busy state, shunting the delay sensitive flow, and after shunting is finished, restoring the routing table to the state of the routing table obtained in the last step (2 b).
Compared with the prior art, the invention has the following advantages:
first, because the present invention calculates the queue occupancy rate of each output link of a node, rather than the entire queue of the node, when dividing the congestion state, the neighbor nodes can know the congestion condition of each link more accurately, and only detours the packets that will pass through the congested link, rather than detours the packets that pass through the corresponding node of the congested link, thereby avoiding the detours of a part of the packets that only pass through the end node of the congested link but not the congested link, and enabling the part of the packets to pass through the optimal path without unnecessary detours, and finally achieving the local load balance of the network.
Secondly, the invention classifies the flow, processes the flow with different service demands, and reduces the end-to-end time delay of the time delay sensitive flow caused by overlong queuing time delay.
Drawings
FIG. 1 is a flow chart of an implementation of the present invention;
FIG. 2 is a sub-flow diagram of congestion handling in the present invention;
FIG. 3 is a schematic diagram of a scenario in which link congestion occurs in a satellite network;
fig. 4 is a schematic diagram of a scenario in which there are two traffic demands of traffic in a satellite network and link congestion occurs.
Detailed Description
Embodiments of the present invention are described in further detail below with reference to the accompanying drawings.
Referring to fig. 1, the overall implementation steps of the present invention are as follows:
step 1, dividing a time slice and obtaining a group of static topological graphs.
And dividing the satellite network into time slices at equal intervals, wherein the time interval between the time slices is the minimum time interval of the structural topology change of the satellite network, and finally obtaining a group of topological graphs at equal time intervals.
And 2, performing route planning on the divided time slices.
2.1) taking propagation delay and queuing delay as real-time link cost measurement, adopting Dijkstra algorithm to calculate the shortest path of each time slice topology, and specifically realizing the method 2.1);
2.2) updating the routing table according to the shortest path, namely writing the next hop corresponding to the shortest path into the next hop field of the routing table to obtain a new routing table.
2.3) waiting for a time interval of one time slice;
2.4) judging whether the ending time is reached, if so, ending the path planning, otherwise, carrying out the processing of 2.1) -2.3) on the next time slice.
And 3, carrying out different treatments on abnormal conditions occurring in the route planning process.
3.1) monitoring the abnormal condition of the link in real time in the process of planning the route;
if the link is congested, different congestion processing is carried out according to the service requirement, and the link is not congested any more;
3.2) judging whether the ending time is reached, if so, stopping the real-time monitoring of the link, otherwise, continuing to execute 3.1).
In the following different embodiments of handling are given for congestion occurring in the monitoring.
Example 1: and (4) routing planning for the traffic with only one service demand in the network and the occurrence of the link congestion condition.
And step A, dividing a time slice and obtaining a group of static topological graphs.
And taking the minimum time interval of the topological change of the satellite network structure as the time interval between each time slice, and dividing the topological partitions of the satellite network at equal intervals in time to finally obtain a group of topological graphs corresponding to each time slice one by one.
And step B, performing route planning on the divided time slices.
B1) Taking propagation delay and queuing delay as real-time link cost measurement, adopting Dijkstra algorithm to calculate the shortest path of the topology of the first time slice:
b1.1) define the real-time link cost metric as follows:
Lcost(t)＝Tprop+Tqueue(t)
wherein L iscost(T) a real-time link cost metric at time T, TpropFor propagation delay of the link, Tqueue(t) is queuing delay of the link at time t;
b1.2) taking the real-time link cost measurement as a path weight;
b1.3) selecting a certain node s in the network, and setting two sets for the node s: a source node set A and a destination node set B;
b1.4) initially, a source node set A only contains a node s, and a destination node set B contains other nodes except s in the network;
b1.5) selecting a node k with the minimum path weight from the destination node set B, adding the node k into the source node set A, and simultaneously removing the node k from the destination node set B;
b1.6) updating the path weight from each node to the node s in the destination node set B: that is, for the case of (s, v) > (s, k) + (k, v), change (s, v) to (s, k) + (k, v), where (s, v) is the path weight from node s to node v, (s, k) is the path weight from node s to node k, and (k, v) is the path weight from node k to node v;
b1.7) repeating the steps (2a5) and (2a6) until all nodes are traversed to obtain the shortest path from the node s to other nodes in the network;
b1.8) executing the processes of (2a3) to (2a7) to all nodes in the network to obtain the shortest paths from all nodes to other nodes in the network;
B2) and updating the routing table of the first time slice according to the shortest path of the first time slice.
And step C, processing the condition that congestion occurs and only one service in the network needs flow during the route planning process.
The satellite network is composed of a series of nodes which are regularly arranged, each node has four output links which are directly connected with four neighbor nodes, each node has a routing table of the node, fig. 3 is a scene schematic diagram when the link congestion occurs in the satellite network, wherein the state of the S45- > S44 link undergoes the process of gradually changing from idle to congestion, and the congestion processing needs to be carried out on the link.
Referring to fig. 2, the processing flow of this step is as follows:
C1) setting up queues at four output links of each satellite node, and calculating the real-time occupancy rate mu (t) of each satellite node according to the length of each queue:
μ(t)＝q(t)÷Qtotal
wherein Q (t) is the queue length at time t, QtotalIs the total length of the queue;
C2) setting an idle threshold value alpha and a congestion threshold value beta, and dividing the congestion state of a satellite output link into three states of idle, busy and busy according to the relation between the real-time occupancy rate [ mu ] (t) of a queue and the idle threshold value alpha and the congestion threshold value beta:
when mu (t) < alpha, dividing the output link state of the direction into an idle state;
when alpha < mu (t) < beta, dividing the output link state of the direction into a busy state;
when mu (t) > beta, dividing the output link state in the direction into a busy state;
the idle threshold value alpha and the congestion threshold value beta are set according to the congestion probability of the link, and the idle threshold value alpha and the congestion threshold value beta meet the condition that alpha is beta/2;
C3) processing for changing the link state from idle to busy:
referring to FIG. 3, the node S45 has four neighboring nodes S44, S35, S46, S55, and the routing table of the node S45, as shown in Table 1.
Routing table of table 1S45
Destination node Next hop node
S33 S44
S34 S35
S43 S44
S44 S44
S54 S44
S65 S55
… …
When the output link queue occupancy rates of S45 to S44 increase to the idle threshold value alpha, the link state of S45- > S44 changes from idle to busy, S45 finds the routing table entry to pass through the link of S45- > S44 from the routing table of the S45, extracts the corresponding destination nodes S33, S43, S44 and S54, and forms a destination node set { S33, S43, S44 and S54 }. S45 encapsulates the destination node set in the main structure of the warning packet, and sends it to the neighboring nodes S35, S46, S55 unrelated to the S45- > S44 link, and at the same time notifies these neighboring nodes that the S45- > S44 link is about to be congested, so that these neighboring nodes find an alternative route not including the S45- > S44 link for the route corresponding to the routing entry whose destination node is S33, S43, S44, S54 in their own routing table, and update it to the routing table, so that the next hop node of the routing entry of the S45- > S44 link that is originally going to pass through a busy state in S35, S46, S55 changes from S45 to S34, S56, S54 in the alternative route;
C4) processing of link status change from busy to busy:
referring to fig. 3, when the output link queue occupancy of S45 to S44 increases to the congestion threshold β, the link status thereof changes from busy to busy, S45 transmits a signaling packet of any one of the structures as a congestion notification packet to the neighboring nodes S35, S46, S55 that have transmitted the warning packet, notifying these neighboring nodes S45->When the S44 link is congested, the neighboring nodes start to use the alternative route planned in the previous stage for shunting, and the routing is reduced to S45->The traffic rate on the link of S44, the neighboring nodes S35, S46, S55 will originally pass through S45->The S44 link sends the traffic to the nodes in the destination node set { S33, S43, S44, S54} to be split according to the split ratio chi, that is, the traffic with the chi ratio still passes through S45->The S44 link is sent to the destination node, the remaining 1-chi fraction of traffic is passed through S35->S34 Link, S46->S56 Link, S55->The S54 link is sent to the destination node,the split ratio
Figure BDA0003501060790000071
Wherein the content of the first and second substances,
Figure BDA0003501060790000072
for shunting the expected traffic of the back-link, IsShunting the flow of the front link;
C5) processing for changing the link state from busy to idle:
referring to fig. 3, when the output link queue occupancy of the node S45 to the node S44 decreases to the idle threshold α, the link status changes from busy to idle, and in order to make the S45- > S44 link have sufficient recovery time, the shunting of C4) continues for a period of time τ, and then stops, and the routing table is restored to the state of the routing table obtained in the last B2).
In this embodiment, packets going through node S45 to S34 that are not associated with the busy S45- > S44 link can still be routed according to the shortest path without the need for re-planning resulting in unnecessary latency increases. For the traffic which passes through the S44- > S45 link and reaches S45, the traffic is not sent back to S44 through the congested link S45- > S44, so the S44 does not need to be notified when the S45 notifies its neighbor, which further reduces the signaling overhead.
Example 2: and (4) routing planning for the flow with two service demands in the network and the occurrence of the link congestion condition.
Step one, dividing a time slice and obtaining a group of static topological graphs.
The specific implementation of this step is the same as step a of the first embodiment.
And step two, performing route planning on the divided time slices.
The specific implementation of this step is the same as step B of the first embodiment.
And step three, processing the situation that congestion occurs and the flow of two services in the network is required during the route planning process.
3a) Setting up queues at four output links of each satellite node, and calculating the real-time occupancy rate mu (t) of each satellite node according to the length of each queue:
the specific implementation of this step is the same as step C1) of the first embodiment.
3b) Setting an idle threshold value alpha and a congestion threshold value beta, and dividing the congestion state of a satellite output link into three states of idle, busy and busy according to the relation between the real-time occupancy rate [ mu ] (t) of a queue and the idle threshold value alpha and the congestion threshold value beta:
the specific implementation of this step is the same as step C2) of the first embodiment.
3c) Dividing the flow in the network into delay sensitive flow and non-delay sensitive flow;
3d) processing delay sensitive traffic and non-delay sensitive traffic of which the link state is changed from an idle state to a busy state:
referring to FIG. 4, each node in the satellite network stores its own routing table, where node S45 has four neighboring nodes S44, S35, S46, S55, and when the output link queue occupancy of S45 to S44 increases to the idle threshold α, the S45- > S44 link state changes from idle to busy, and Table 1 gives the routing table for node S45 at this time.
The delay sensitive traffic and the non-delay sensitive traffic in the network are processed as follows:
from the routing table in table 1, the S45 node finds a routing table entry to be passed through the S45- > S44 link and extracts its corresponding destination nodes S33, S43, S44, S54, and forms a destination node set { S33, S43, S44, S54 }. The S45 encapsulates the destination node set in a warning packet, and sends it to the neighboring nodes S35, S46, S55 that are not related to the S45- > S44 link, notifies these neighboring nodes S45- > S44 links of imminent congestion, and makes them look for two alternative routes that do not include the S45- > S44 link for the routing table entries of the destination nodes S33, S43, S44, S54, and update them into the routing table, so that the next hop node of the routing table entry that originally passes through the S45- > S44 link in a busy state in S35, S46, S55 becomes S34, S56, S54 in the alternative route 1, and becomes S25, S36, S65 in the alternative route 2;
3e) processing delay sensitive traffic and non-delay sensitive traffic of which the link state is changed from a busy state to a busy state:
3e1) detouring the non-delay sensitive flow, and sending the delay sensitive flow according to the original path, namely sending the non-delay sensitive flow to a congestion link according to the set flow splitting ratio x by the corresponding neighbor node, and transmitting the rest 1-x ratio non-delay sensitive flow through the found alternative route 1;
3e2) if a period of time tau has elapsed0If the back link is still in a busy state, shunting the delay sensitive traffic, namely forwarding the delay sensitive traffic with the ratio of 1-chi to an alternative route 1, and forwarding the non-delay sensitive traffic to an alternative route 2;
3f) processing delay sensitive traffic and non-delay sensitive traffic of which the link state is changed from a busy state to an idle state:
referring to fig. 4, when the occupancy rate of the output link queue of S45 to S44 decreases to the idle threshold α, the link status changes from busy to idle, and in order to make the S45- > S44 link have enough recovery time, the shunting of 3e) continues for a period of time τ, and then stops, and the routing table recovers to the state of the routing table obtained in the previous step two.
In this embodiment, when the flow is split, the non-delay sensitive traffic is preferentially split, if the S45- > S44 link is still congested after the non-delay sensitive traffic completely bypasses, another two selectable paths are planned for the routing table entry in the neighboring node to the destination node centralized node, the delay sensitive traffic with the χ ratio is still sent to the destination node through the S45- > S44 link according to the original path, the remaining traffic with the 1- χ ratio is forwarded through the suboptimal path, and all the non-delay sensitive traffic is forwarded through the third selectable path.
The above are two specific examples of the present invention, and do not constitute any limitation to the present invention, and all modifications and changes made within the spirit and scope of the present invention shall fall within the protection scope of the present invention.

Claims (10)

1. A load balancing routing method based on accurate link state feedback is characterized by comprising the following steps:
(1) dividing the satellite network into equally spaced time slices in time to turn the high-speed dynamic masking of satellite movement into a series of static topologies;
(2) and (3) carrying out route planning on the time slices divided in the step (1):
(2a) taking propagation delay and queuing delay as real-time link cost measurement, and adopting Dijkstra algorithm to calculate the shortest path of each time slice;
(2b) updating the routing table according to the shortest path, namely writing the next hop corresponding to the shortest path into the next hop field of the routing table to obtain a new routing table;
(2c) monitoring the link state in real time: if the link congestion occurs, executing (3); otherwise, continuing monitoring;
(3) and (3) shunting the condition that the link is congested:
(3a) setting up queues at four output links of each satellite node, calculating real-time occupancy rates mu (t) of the queues, and dividing congestion states of the satellite output links into three states of idle, busy and busy according to the occupancy conditions of the queues;
(3b) when a certain output link is changed from idle to busy, the satellite searches a routing table of the satellite to find all routing table items of which the next hop passes through the output link, and sends a destination node set corresponding to the table items to all other neighbor nodes except the neighbor node corresponding to the output link of the node through a warning packet; after receiving the warning packet, the neighbor node searches for a substitute route which does not comprise a congestion link for the packet to the node in the destination node set, and updates the substitute route into a route table;
(3c) when a certain output link is changed from a busy state to a busy state, the satellite sends a congestion notification packet to a neighbor node which has sent a warning packet, so that the corresponding neighbor node sends the flow to the congestion link according to the set flow dividing ratio x, and transmits the residual flow with the ratio of 1-x through the found alternative route;
(3d) when a certain output link is changed into an idle state from a busy state, after the shunting of the step (3b) continues to be continued for a period of time tau, the routing table is restored to the state of the routing table obtained in the last step (2 b);
(3e) if different types of service requirements exist in the network, dividing the flow in the network into delay sensitive flow and non-delay sensitive flow, bypassing the non-delay sensitive flow, if the link is still in a busy state, shunting the delay sensitive flow, and after shunting is finished, restoring the routing table to the state of the routing table obtained in the last step (2 b).
2. The method of claim 1, wherein the interval of time slices divided into for the satellite network in (1) is a minimum time interval of topology change of the satellite network structure.
3. The method of claim 1, wherein the Dijkstra algorithm is used in (2a) to calculate the shortest path for the topology of each time slice, and the following is implemented:
(2a1) the real-time link cost metric is defined as follows:
Lcost(t)＝Tprop+Tqueue(t)
wherein L iscost(T) a real-time link cost metric at time T, TpropFor propagation delay of the link, TqueueAnd (t) is the queuing delay of the link at the time t.
(2a2) Taking the real-time link cost measurement as a path weight;
(2a3) selecting a certain node s in the network, and setting two sets for the node s: a source node set A and a destination node set B;
(2a4) initially, a source node set A only contains a node s, and a destination node set B contains other nodes except the node s in the network;
(2a5) selecting a node k with the minimum path weight from the destination node set B, adding the node k into the source node set A, and simultaneously removing the node k from the destination node set B;
(2a6) and updating the path weight from each node to the node s in the destination node set B: that is, for the case of (s, v) > (s, k) + (k, v), change (s, v) to (s, k) + (k, v), where (s, v) is the path weight from node s to node v, (s, k) is the path weight from node s to node k, and (k, v) is the path weight from node k to node v;
(2a7) repeating the steps (2a5) and (2a6) until all nodes are traversed to obtain the shortest path from the node s to other nodes in the network;
(2a8) and (2) executing the processes of (2a3) to (2a7) on all nodes in the network to obtain the shortest paths from all nodes in the network to other nodes.
4. The method of claim 1, wherein the queue real-time occupancy μ (t) is calculated in (3a) as follows:
μ(t)＝q(t)÷Qtotal
where Q (t) is the queue length at time t, QtotalIs the total length of the queue.
5. The method according to claim 1, wherein the division of the state of the satellite output link according to the queue occupancy in (3a) is performed according to the relationship between the queue real-time occupancy μ (t) and the set idle threshold α and congestion threshold β:
when mu (t) < alpha, dividing the output link state in the direction into an idle state;
when alpha is less than mu (t) and less than beta, dividing the output link state in the direction into a busy state;
when mu (t) > beta, dividing the output link state in the direction into a busy state;
wherein, the idle threshold value alpha and the congestion threshold value beta are set according to the congestion probability of the link, and the idle threshold value alpha and the congestion threshold value beta meet the following conditions: α ═ β/2.
6. The method of claim 1, wherein the alert packet in (3b) is composed of a body structure and other structures, wherein the body structure is a destination node set, and the other structures are optional.
7. The method of claim 1, wherein the congestion notification packet in (3c) is a signaling packet with any structure.
8. The method according to claim 1, wherein the split ratio χ in (3c) is calculated as follows:
Figure FDA0003501060780000031
wherein the content of the first and second substances,
Figure FDA0003501060780000032
for shunting the expected traffic of the back-link, IsThe traffic of the front link is shunted.
9. The method of claim 1, wherein the non-delay-sensitive traffic is bypassed in (3e), and the corresponding neighbor node sends the non-delay-sensitive traffic to the congested link according to the set splitting ratio χ, and transmits the remaining non-delay-sensitive traffic with 1- χ ratio through the found alternative route.
10. The method according to claim 1, wherein the splitting of the delay-sensitive traffic in (3e) is to forward the delay-sensitive traffic with a ratio of 1- χ to the secondary selected path and to forward the non-delay-sensitive traffic to the third optimal path except for the primary selected path and the secondary selected path.
CN202210127411.1A 2022-02-11 2022-02-11 Load balancing routing method based on accurate link state feedback Pending CN114499644A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210127411.1A CN114499644A (en) 2022-02-11 2022-02-11 Load balancing routing method based on accurate link state feedback

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210127411.1A CN114499644A (en) 2022-02-11 2022-02-11 Load balancing routing method based on accurate link state feedback

Publications (1)

Publication Number Publication Date
CN114499644A true CN114499644A (en) 2022-05-13

Family

ID=81479079

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210127411.1A Pending CN114499644A (en) 2022-02-11 2022-02-11 Load balancing routing method based on accurate link state feedback

Country Status (1)

Country Link
CN (1) CN114499644A (en)

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20070201371A1 (en) * 2006-02-28 2007-08-30 Microsoft Corporation Congestion adaptive network data routing
CN105227483A (en) * 2015-08-21 2016-01-06 北京空间飞行器总体设计部 Based on the low complex degree Load Balance Routing Algorithms of LEO satellite network
CN110290066A (en) * 2019-06-24 2019-09-27 中国空间技术研究院 Satellite network dynamic routing method based on queue monitoring and congestion prediction
CN111148161A (en) * 2020-01-06 2020-05-12 中国科学院微小卫星创新研究院 Method and system for balancing load route between low-orbit satellite constellation satellites
CN111416655A (en) * 2020-04-07 2020-07-14 南京邮电大学 Low-orbit satellite routing improvement method based on virtual topology
CN113507722A (en) * 2021-07-05 2021-10-15 北京邮电大学 Implementation method of NS 3-based platform for controlling congestion of low-orbit satellite

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20070201371A1 (en) * 2006-02-28 2007-08-30 Microsoft Corporation Congestion adaptive network data routing
CN105227483A (en) * 2015-08-21 2016-01-06 北京空间飞行器总体设计部 Based on the low complex degree Load Balance Routing Algorithms of LEO satellite network
CN110290066A (en) * 2019-06-24 2019-09-27 中国空间技术研究院 Satellite network dynamic routing method based on queue monitoring and congestion prediction
CN111148161A (en) * 2020-01-06 2020-05-12 中国科学院微小卫星创新研究院 Method and system for balancing load route between low-orbit satellite constellation satellites
CN111416655A (en) * 2020-04-07 2020-07-14 南京邮电大学 Low-orbit satellite routing improvement method based on virtual topology
CN113507722A (en) * 2021-07-05 2021-10-15 北京邮电大学 Implementation method of NS 3-based platform for controlling congestion of low-orbit satellite

Similar Documents

Publication Publication Date Title
CN108289064B (en) Hybrid load balancing method in data center network
US5600638A (en) Method and system for improving the processing time of the path selection in a high speed packet switching network
JP2964957B2 (en) High-speed routing control method
JP2679895B2 (en) Information transmission method and device
CN107196854B (en) Data plane exception handling method in software defined network
Li et al. A state-aware and load-balanced routing model for LEO satellite networks
JPH0766835A (en) Communication network and method for selection of route in said network
GB2303996A (en) Communications arrangement and method
CN112350949B (en) Rerouting congestion control method and system based on flow scheduling in software defined network
CN108111411B (en) Backbone network and dynamic path planning system and planning method thereof
CN112311441B (en) Congestion avoidance routing method in low-orbit constellation network
CN110708736A (en) Dynamic routing method and system based on energy efficiency relay selection
CN112532518A (en) Path selection method and device of segment routing strategy
CN105978805A (en) Anycast attribute service oriented energy-saving routing method
CN105472484A (en) Wave channel balancing route wavelength allocation method of power backbone optical transport network
CN114286382A (en) Capacity-breaking reconfigurable routing strategy based on prior knowledge base
CN114499644A (en) Load balancing routing method based on accurate link state feedback
Hertiana et al. A joint approach to multipath routing and rate adaptation for congestion control in openflow software defined network
Huang et al. An adaptive multipath routing for LEO satellite network
CN113852415B (en) Global flow scheduling method suitable for low-orbit satellite network
JP3563328B2 (en) Route control system
CN110392408B (en) Time-varying network topology control method and system based on node role and adjustable parameters
US7609965B2 (en) Method for reorganizing wavelength division multiplexing network using dynamic watermarks
Zhong et al. Research and Implementation of AOMDV Multipath Routing Protocol
Beaubrun et al. A routing algorithm for distributed communication networks

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
WD01 Invention patent application deemed withdrawn after publication
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20220513