An Adaptive Routing Method and System for Multilayer Networks Based on Stability Constraints

Abstract

This invention provides a multi-layer network adaptive routing method and system based on stability constraints, relating to the field of adaptive routing technology. The method includes dividing multiple wireless networks into communication areas to obtain a multi-layer topology representation; for multiple communication node pairs in the multi-layer topology representation, statistically analyzing historical transmission behavior during data forwarding to obtain link historical state vectors and path end-to-end historical state vectors; calculating link comprehensive evaluation indicators and path end-to-end success rates based on the link historical state vectors and path end-to-end historical state vectors; using the link comprehensive evaluation indicators to screen communication areas for stability, constructing a candidate area-level path set; and performing adaptive routing selection for multi-armed gambling machines based on the candidate area-level path set and path end-to-end success rates to obtain an adaptive routing path for each communication node pair. This invention solves the problem of routing oscillation caused by frequent switching of transmission paths in existing methods.

Description

Technical Field

[0001] This invention relates to the field of adaptive routing technology, and more specifically, to an adaptive routing method and system for multilayer networks based on stability constraints. Background Technology

[0002] With the continuous expansion of wireless network scale and the dynamic changes in network topology, adaptive routing technology is widely used in multi-wireless network cooperative communication scenarios. Existing adaptive routing methods typically use instantaneous measurement indicators such as link latency, bandwidth occupancy, queue length, or channel utilization as the basis for routing costs and select paths based on the current network state. However, when wireless networks experience bursts of traffic, short-term interference, or fluctuations in link quality, these instantaneous indicators are prone to significant changes, leading to frequent updates to routing costs. This causes data transmission paths to repeatedly switch between multiple candidate paths, resulting in routing oscillations, increased control overhead, and reduced network stability.

[0003] Furthermore, existing methods typically focus on current optimality while lacking mechanisms to evaluate the long-term stability of links or paths. Even if a path has low cost in the short term, its transmission success rate may fluctuate significantly, leading to packet loss or retransmissions in subsequent transmissions, thus affecting overall communication quality. Many adaptive routes select paths based solely on single-node or local neighborhood information, failing to comprehensively consider the end-to-end transmission effect of the entire path. This can easily result in locally optimal but globally suboptimal routing outcomes, leading to uneven network load distribution and long-term congestion on some links. Some methods attempt to introduce learning or probabilistic decision-making mechanisms to improve routing adaptability, but they typically select paths directly from the entire candidate path space without introducing explicit stability constraints to pre-screen unstable links or regions. This makes the learning process susceptible to interference from unstable factors, making it difficult to achieve long-term stable and reliable routing decisions in dynamic network environments. Summary of the Invention

[0004] The purpose of this invention is to provide an adaptive routing method and system for multi-layer networks based on stability constraints, in order to improve the aforementioned problems. To achieve the above objective, the technical solution adopted by this invention is as follows:

[0005] Firstly, this application provides an adaptive routing method for multi-layer networks based on stability constraints, comprising:

[0006] Multiple wireless networks are divided into communication areas to obtain a multi-layer topology representation that includes multiple communication areas, each of which includes multiple wireless network nodes.

[0007] For multiple communication node pairs with communication needs in a multi-layer topology representation, historical transmission behavior is statistically analyzed during data forwarding to obtain the link historical state vector and the path end-to-end historical state vector.

[0008] Based on the link historical state vector and the path end-to-end historical state vector, calculate the link comprehensive evaluation index and the path end-to-end success index respectively.

[0009] The stability of the communication area is screened by link comprehensive evaluation index, and a set of candidate area-level paths that meet the stability constraints is constructed.

[0010] Based on the candidate regional path set and the path end-to-end success rate index, adaptive routing selection is performed for the multi-armed gambling machine to obtain the adaptive routing path for each communication node pair.

[0011] Secondly, this application also provides a multi-layer network adaptive routing system based on stability constraints, comprising:

[0012] The partitioning module is used to divide multiple wireless networks into communication areas to obtain a multi-layer topology representation that includes multiple communication areas, each of which includes multiple wireless network nodes.

[0013] The statistics module is used to collect historical transmission behavior data during the data forwarding process for multiple communication node pairs with communication needs in a multi-layer topology representation, and obtain the link historical state vector and the path end-to-end historical state vector.

[0014] The calculation module is used to calculate the comprehensive evaluation index of the link and the end-to-end success index of the path based on the historical state vector of the link and the historical state vector of the path end-to-end.

[0015] The filtering module is used to filter communication areas for stability based on comprehensive link evaluation indicators and construct a set of candidate area-level paths that meet stability constraints.

[0016] The routing module is used to perform adaptive routing selection for multi-armed gambling machines based on the candidate regional path set and the path end-to-end success rate index, so as to obtain the adaptive routing path for each communication node pair.

[0017] The beneficial effects of this invention are as follows:

[0018] (1) This invention constructs a multi-layer topology representation including node level and region level, and constructs evaluation indicators in combination with historical transmission behavior, thereby screening unstable communication areas, eliminating highly volatile and unreliable routing paths, avoiding frequent path switching caused by instantaneous indicator fluctuations, and ensuring the continuity and stability of data transmission. At the same time, the long-term communication effect of the entire routing path is quantified by the end-to-end success index of the path, which makes up for the shortcomings of traditional methods that only focus on the local link status;

[0019] (2) The present invention also introduces a multi-armed gambling machine to make adaptive routing decisions, maps the regional transmission path after stability screening to candidate arms, and dynamically updates it based on historical benefits and real-time feedback. While making full use of high-quality path resources, it retains the ability to explore potential better paths, adapts to the scenario requirements of dynamic changes in wireless network topology and fluctuations in service traffic, and avoids routing oscillations caused by frequent switching of transmission paths when service load changes.

[0020] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing embodiments of the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description, claims, and drawings. Attached Figure Description

[0021] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a schematic diagram of the adaptive routing method for multilayer networks based on stability constraints as described in an embodiment of the present invention.

[0023] Figure 2 This is a schematic diagram of the structure of a multi-layer network adaptive routing system based on stability constraints as described in an embodiment of the present invention. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0025] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this invention, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0026] Example 1:

[0027] This embodiment provides an adaptive routing method for multilayer networks based on stability constraints.

[0028] It should be noted that the communication network comprises multiple heterogeneous wireless networks, which can include various types such as cellular communication networks, wireless ad hoc networks, and dedicated wireless networks. Furthermore, these wireless network nodes are distributed across different geographical areas and form a multi-hop communication topology through wireless links. Due to differences in access methods, transmission rates, coverage, and interference resistance among different wireless networks, as well as variations in node density, wireless environment, and service load, link stability and communication quality differ significantly across regions. The overall network exhibits regionalized, hierarchical, and unevenly stable topological characteristics.

[0029] In this application scenario, the network simultaneously contains both periodic and bursty service data. Service traffic changes dynamically over time, and some wireless links are susceptible to quality fluctuations due to environmental interference, node movement, or channel contention. When the data transmission path crosses multiple communication areas or includes multi-hop wireless links, fluctuations in the state of local links are easily amplified during end-to-end transmission, leading to overall path performance instability.

[0030] If existing adaptive routing methods based on instantaneous link cost are used, transmission paths are easily switched frequently when link status fluctuates or service load changes, leading to routing oscillations. Routing oscillations not only increase network control signaling and computational overhead, but may also cause out-of-order packets, increased retransmissions, and end-to-end latency fluctuations, thereby reducing the stability and success rate of end-to-end transmission and affecting the overall quality of communication services.

[0031] See Figure 1 The figure shows that the method includes steps S1, S2, S3, S4 and S5.

[0032] Step S1: Divide the multiple wireless networks into communication areas to obtain a multi-layer topology representation including multiple communication areas, where each communication area includes multiple wireless network nodes;

[0033] Step S1 includes:

[0034] Step S11: Obtain basic network information of multiple wireless networks, including identification information, communication relationships, link communication capability parameters, and location parameters of wireless network nodes;

[0035] In this step, the link communication capability parameters include, but are not limited to, link bandwidth, average latency, packet loss rate, bit error rate, signal strength, or signal-to-noise ratio.

[0036] Step S12: Based on basic network information, construct a node-level network topology with wireless network nodes as vertices and communicable links as edges;

[0037] Step S13: According to the preset communication area division rules, the wireless network nodes in the node-level network topology diagram are divided into multiple communication areas.

[0038] In this step, the rules for dividing the communication area can be constructed based on the geographical distance of the nodes, the link connection density or connectivity, or the network management strategy.

[0039] Step S14: For each communication area, the wireless network node in that communication area that has a communication connection with other communication areas is designated as the area interface node;

[0040] Understandably, within a communication area, not every wireless network node can directly communicate with other communication areas across different areas. These nodes, which have direct communication links with other communication areas, determine the reachability and stability of subsequent cross-area paths, while also avoiding ineffective cross-area path searches on nodes within the same area.

[0041] For a communication region, identify its set of regional interface nodes:

[0042] ;

[0043] In the formula, express The set of regional interface nodes, Represents a wireless network node , Indicates communication area , Represents a wireless network node , Indicates communication area , Represents a set of links.

[0044] in, express and The link formed, i.e. and Direct communication is possible.

[0045] Step S15: Abstract each communication area into a region-level node, and construct a multi-layer topology representation including node-level network topology and region-level network topology based on the communication connection relationship between region interface nodes.

[0046] In this step, the multi-layer topology actually refers to a multi-layer network, which is reflected in the regional level (regional network topology of the communication area) and the intra-regional level (node-level network topology within the communication area).

[0047] Step S2: For multiple communication node pairs with communication needs in the multi-layer topology representation, statistical analysis of historical transmission behavior is performed during data forwarding to obtain the link historical state vector and the path end-to-end historical state vector.

[0048] In step S2, obtaining the link historical state vector and the path end-to-end historical state vector includes:

[0049] Step S21: Statistically analyze the historical transmission behavior information of all communication links in the multi-layer topology representation. The historical transmission behavior information includes the number of successful forwardings, the number of failures, the number of timeouts, the number of retransmissions, and the duration of continuous stable transmission.

[0050] It is understandable that the result of a single transmission is accidental, and only long-term statistical behavior can reflect whether the link is reliable. In order to avoid short-term congestion or occasional interference misleading routing decisions, this embodiment continuously monitors each communication link in the multi-layer topology representation and counts its historical transmission behavior information within a preset statistical time window, that is, counts the number of successful forwards. Number of failures Number of timeouts Number of retransmissions and the duration of continuous stable transmission .

[0051] The duration of continuous stable transmission represents the cumulative time during which the link maintains a stable transmission state without failure, timeout, or retransmission.

[0052] Step S22: Construct a link history state vector for each communication link based on the historical transmission behavior information of the communication link. The communication link is a link formed by a pair of wireless network nodes that can directly communicate wirelessly.

[0053] Step S23: In the multi-layer topology representation, identify communication node pairs that have data communication needs;

[0054] In this step, in the multi-layer topology representation, based on the service request or data flow requirements, the communication node pairs that need to perform end-to-end data transmission are identified. Each communication node pair consists of a source node and a corresponding target node, and the communication node pairs can be triggered by the service scheduling module, application layer requests, or network management policies.

[0055] Step S24: For each communication node pair, obtain the end-to-end transmission path actually used in the data forwarding process, wherein the end-to-end transmission path includes multiple wireless network nodes passed through in sequence;

[0056] It is understandable that a communication link is communication between two nodes, but whether end-to-end is successful is the result of the overall behavior of the path. Relying solely on the communication link cannot reflect the cumulative effect of path-level risks. Therefore, it is necessary to obtain the actual end-to-end transmission path used during data forwarding.

[0057] Specifically, for each communication node pair During data forwarding, the actual end-to-end transmission path used is recorded. Besides the source node and its corresponding destination node, the end-to-end transmission path also includes multiple intermediate nodes (hop-by-hop forwarding nodes). Furthermore, for each pair of communication nodes, there are multiple end-to-end transmission paths. and They represent the first A pair of communication nodes, consisting of a source node and a target node.

[0058] Step S25: Construct the path end-to-end historical state vector of the corresponding communication node pair by using the historical transmission behavior information of multiple end-to-end transmission paths of the same communication node pair.

[0059] In this step, a corresponding path history state vector is constructed for each end-to-end transmission path. The corresponding path end-to-end history state vector is obtained by using the path history state vectors of all the paths under a communication node.

[0060] Step S3: Based on the link historical state vector and the path end-to-end historical state vector, calculate the link comprehensive evaluation index and the path end-to-end success index respectively;

[0061] In this step, the link comprehensive evaluation index can characterize the long-term stability and reliability of a single communication link, while the path end-to-end success index is used to characterize the overall success level of the end-to-end transmission path in the historical communication process.

[0062] Specifically, the link historical state vector is :

[0063] ;

[0064] In the formula, Indicates the first The link history state vector of each communication link. Indicates the first The number of successful forwards on each communication link Indicates the first The number of forwarding failures on each communication link. Indicates the first The number of timeouts for each communication link. Indicates the first The number of retransmissions on each communication link. Indicates the first The duration of continuous and stable transmission of a communication link.

[0065] Calculate the comprehensive evaluation index of the link by using the link's historical state vector. :

[0066] ;

[0067] In the formula, Indicates the first Comprehensive evaluation indicators for communication links. , and All of these represent weighting coefficients used to balance success rate, cost penalty, and stability contribution. Indicates the first The length of the historical statistical time window for each communication link. Indicates the first The number of successful forwards on each communication link Indicates the first The number of forwarding failures on each communication link. Indicates the first The number of timeouts for each communication link. Indicates the first The number of retransmissions on each communication link. Indicates the first The duration of continuous and stable transmission of a communication link.

[0068] The evaluation criteria are as follows: the first is the success rate, reflecting the long-term availability of the link; the second is the retransmission penalty, reflecting the implicit instability risk of the link; and the third is the stability and sustainability, reflecting the link's resilience to fluctuations. By calculating these comprehensive evaluation metrics, the communication link can be better evaluated than it may be due to short-term high throughput or occasional successes.

[0069] In this step, since the success of end-to-end communication is the result of the cumulative effect of multiple hops, even if a single link performs stably, an anomaly at any node or link in the path can lead to overall transmission failure. Therefore, relying solely on link-level evaluation cannot accurately reflect the risks of end-to-end communication. This embodiment calculates the path end-to-end success rate index for each end-to-end transmission path, enabling subsequent path selection to be optimized directly based on historical overall communication performance.

[0070] Specifically, for the same pair of communication nodes, the path history state vector of each end-to-end transmission path is obtained through the path end-to-end historical state vector. :

[0071] ;

[0072] In the formula, express The path history state vector, express The number of successful forwards, express The number of failed forwarding attempts. express The number of timeouts, express The number of retransmissions, express The duration of continuous and stable transmission, Indicates communication node pair The An end-to-end transmission path.

[0073] The path end-to-end success rate index is calculated by using the path history state vector. :

[0074] ;

[0075] In the formula, express The end-to-end success rate metric for the path. , and Both represent path-level weight parameters. express Path statistics time window express The number of successful forwards, express The number of failed forwarding attempts. express The number of timeouts, express The number of retransmissions, express The duration of continuous and stable transmission, Indicates communication node pair The An end-to-end transmission path.

[0076] Step S4: Use comprehensive link evaluation metrics to screen communication areas for stability and construct a set of candidate area-level paths that meet stability constraints;

[0077] In this step, a regional stability constraint mechanism is introduced before path selection. The long-term stability evaluation results of the link layer are mapped to the regional layer, and the transmission paths corresponding to unstable regions are actively eliminated in the regional network topology. This effectively reduces the negative impact of unstable links and highly volatile regions on the success rate of end-to-end communication, and provides a stable and reliable candidate path space for the subsequent adaptive routing selection of the multi-armed gambling machine.

[0078] Step S4 includes:

[0079] Step S41: For each communication region in the multi-layer topology representation, obtain the comprehensive evaluation index of all communication links within that communication region;

[0080] Step S42: Calculate the regional stability index of the communication area based on the comprehensive evaluation index of all links within the communication area;

[0081] Because link states in wireless networks are spatially highly correlated, communication links within the same communication area are often simultaneously affected by similar environmental interference and changes in service load. By aggregating the comprehensive evaluation indicators of multiple communication links within a communication area, the interference of single communication link anomalies on decision-making can be effectively reduced, allowing the regional stability indicators to reflect the long-term communication reliability level of the communication area.

[0082] Secondly, when some links within a region exhibit significant instability, even if other links perform well in the short term, the region as a whole still faces a high communication risk. To avoid using only the average value, which could mask risks with a few high-quality links, and to avoid using only the minimum value, which could be overly sensitive to occasional anomalies, this step adopts a regional stability index with consistency penalties. This introduces constraints on inconsistencies within the region while maintaining an overall stability assessment.

[0083] Specifically, calculation The average comprehensive evaluation index of the link:

[0084] ;

[0085] In the formula, express The average of the comprehensive evaluation indicators of the link, Indicates communication area , express The number of communication links, Indicates the first Comprehensive evaluation indicators for each communication link.

[0086] Define the dispersion of the comprehensive evaluation index of links within a communication area to measure the consistency of link status within the area:

[0087] ;

[0088] In the formula, express The dispersion of the comprehensive evaluation index of the internal link express The number of communication links, Indicates the first Comprehensive evaluation indicators for communication links. express The average of the comprehensive evaluation indicators of the link.

[0089] in, The larger the value, the greater the difference in link stability within the region.

[0090] A consistency penalty function is introduced to penalize regions where link stability distribution is inconsistent:

[0091] ;

[0092] In the formula, express The penalty function, Represents an exponential function. This represents the consistency penalty coefficient. express The degree of dispersion of the comprehensive evaluation index of the internal link.

[0093] Among them, the consistency penalty coefficient This is used to adjust the penalty intensity when the link stability within the region is highly consistent. ,but When there is a clearly unstable link, Increase Rapidly decrease.

[0094] Therefore, regional stability indicators for:

[0095] ;

[0096] In the formula, express Regional stability indicators, express The average of the comprehensive evaluation indicators of the link, express The penalty function.

[0097] Step S43: Compare the regional stability index of the communication area with the preset stability threshold to determine whether the communication area meets the stability constraints;

[0098] In this step, the stability threshold This stability threshold can be preset based on network management policies, historical statistics, or business reliability requirements. It does not aim to select the optimal region, but rather to identify and exclude communication regions that are in a state of high fluctuation over a long period of time and are difficult to provide stable communication quality. Therefore, when If the signal is positive, the communication area is considered stable; otherwise, it is considered unstable.

[0099] Step S44: Using the source and target regions corresponding to the communication node pairs as the starting and ending regions, obtain multiple region-level transmission paths in the multi-layer topology representation;

[0100] In this step, the communication area to which the source node belongs is determined as the source area through the communication node pair. and the communication area to which the target node belongs. .

[0101] Then, in the regional network topology in the multi-layer topology representation, with As the starting region, For the target area, obtain multiple regional-level transmission paths, and each regional-level transmission path is composed of multiple regional-level nodes in sequence.

[0102] Step S45: Eliminate regional-level transmission paths that contain communication areas that do not meet stability constraints to obtain a set of candidate regional-level paths.

[0103] In this step, if any regional node in the regional transmission path is a communication area that does not meet the stability constraints, then the regional transmission path will be removed.

[0104] Step S5: Based on the candidate regional path set and the path end-to-end success rate index, perform adaptive routing selection for the multi-armed gambling machine to obtain the adaptive routing path for each communication node pair.

[0105] It should be noted that Multi-Armed Brawler (MAB) is a classic online reinforcement learning and probabilistic decision-making method, widely used in dynamic decision-making scenarios that require balancing exploration (exploring potential better options) and exploitation (utilizing currently known optimal options). In this embodiment, considering the stability constraints of multi-layer networks and the actual communication characteristics of wireless networks, targeted optimizations were made to multiple aspects, including candidate arm construction, payoff definition, selection strategy, feedback mechanism, and parameter updates. This adapts to the dynamic needs of wireless network scenarios, addresses the pain points of basic MAB in routing applications, and achieves adaptive routing decisions that stably utilize high-quality paths, rationally explore potential paths, and dynamically balance network load.

[0106] Step S5 includes:

[0107] Step S51: For each pair of communication nodes, obtain the corresponding candidate region-level path set;

[0108] Step S52: Construct multiple candidate arms based on the regional transmission paths in the candidate regional path set. The candidate arms include the access node and the exit node of each communication region in the regional transmission path.

[0109] In this step, each candidate regional-level transmission path under a communication node pair is mapped to a candidate arm, and the candidate arm is jointly determined by the sequence of access nodes and exit nodes corresponding to each communication region in the regional-level transmission path. Therefore, a candidate arm uniquely corresponds to a regional-level abstract representation of an end-to-end transmission path, avoiding dimensionality explosion caused by treating single-hop links as arms.

[0110] Step S53: Calculate the historical revenue of the corresponding candidate arm by using the path end-to-end success rate index corresponding to multiple end-to-end transmission behaviors under the same regional transmission path;

[0111] In this step, since the candidate arm only constrains the communication area sequence, access node, and exit node, but does not limit the specific forwarding nodes within the communication area, multiple different end-to-end transmission paths will be generated under the same candidate arm. Therefore, the historical performance of the candidate arm is the combined historical performance of all end-to-end transmission paths under that area-level transmission path.

[0112] Specifically, let candidate arms be defined. Indicates communication node pair The A regional transmission path, in the candidate arm Below, there are multiple end-to-end transmission paths, which will connect the candidate arms. Each end-to-end transmission path is defined as a candidate arm. The candidate paths are listed below, and each candidate path has a corresponding end-to-end success rate metric.

[0113] The excellence of a candidate arm is not solely determined by the performance of a single path, but rather by the overall stability and success of end-to-end communication within the communication region under the constraints of regional path selection. Therefore, defining historical returns based on a success-weighted average weighted by selection frequency reflects the average communication capability of the candidate arm, while reducing sensitivity to extremely good or bad paths under occasional conditions.

[0114] ;

[0115] In the formula, Indicates the historical returns of the candidate arm. This indicates the number of candidate paths under a candidate arm. Indicates the first The number of times each candidate path has been selected in history. Represents the smoothing factor. Indicates the first The end-to-end success rate metric for each candidate path.

[0116] Step S54: Initialize the state parameters of the multi-armed gambling machine using the historical returns and historical selection counts of the candidate arms;

[0117] In this step, the state parameters include the estimated revenue and the number of historical selections for each candidate arm. During initialization, the estimated revenue of a candidate arm is the historical revenue calculated in step S53, and the number of historical selections for a candidate arm is the cumulative number of historical selections for all candidate paths under that candidate arm.

[0118] Step S55: Based on the path selection strategy of the multi-armed gambling machine, candidate arms are selected according to the state parameters to obtain the adaptive routing path of the corresponding communication node pair.

[0119] Step S55 includes:

[0120] Step S551: Based on the path selection strategy of the multi-armed gambling machine, select candidate arms according to the state parameters to obtain the regional routing path;

[0121] In this step, the UCB strategy is adopted, and the candidate arm scoring function is defined:

[0122] ;

[0123] In the formula, Indicates time Candidate arm Choose a rating, Indicates time Candidate arm The estimated revenue, Indicates the exploratory adjustment coefficient. Indicates candidate arm Deadline The number of times the historical selection was made, among which, The actual representation of the current communication node pair Total number of routing decisions made. Used to balance historical gains with exploration progress.

[0124] Choose to The largest candidate arm is the regional routing path for the corresponding communication node pair.

[0125] Step S552: Determine the routing nodes for each communication area in the regional routing path to obtain the adaptive routing path for the corresponding communication node pair;

[0126] In this step, since the regional routing path has already determined the communication area, and the access node and exit node of each communication area have also been determined, it is also necessary to determine the routing node in the middle of each communication area.

[0127] Local routing strategies, such as shortest hop count and maximizing local link evaluation, are employed in each communication area to determine the routing nodes in each communication area. Once all nodes are obtained, the adaptive routing path for the communication node pairs is obtained.

[0128] Step S553: ​​After an end-to-end data transmission is completed on the adaptive routing path, generate the corresponding real-time feedback value based on the actual transmission result;

[0129] In this step, after completing an end-to-end data transmission, the original real-time feedback value is obtained. A value of 1 indicates successful transmission, while a value of 0 indicates transmission failure.

[0130] It is understandable that multiple communication node pairs typically exist simultaneously in a network, and each pair selects its path based on its historical transmission performance. In the absence of coordination or constraint mechanisms, multiple communication node pairs may simultaneously favor the same or a few historically performing paths within a short period. This can lead to multiple communication node pairs simultaneously choosing to traverse the same communication area or the same intermediate nodes, causing a sharp increase in traffic volume in that area or node within a short time. Furthermore, even if this path has historically been stable, when the instantaneous load exceeds the capacity of the area or link, it can still cause increased queuing delays, increased retransmissions, and timeout failures. Multiple nodes may cluster around the same preferred path, thus amplifying the risk of congestion.

[0131] Therefore, in scenarios where multiple communication nodes concurrently select paths, relying solely on historical success rates or static link evaluations is insufficient to effectively suppress the dynamic formation of congestion. This embodiment introduces a congestion-aware penalty mechanism to correct the original instantaneous feedback value. Specifically, when the load of the communication area traversed by a path increases, the instantaneous feedback of that path is automatically reduced, thereby suppressing the probability of it being repeatedly selected in subsequent decisions.

[0132] Specifically, by statistically analyzing the traffic volume of each communication area in the adaptive routing path within the current time window, the current area-level load of each communication area in the adaptive routing path is calculated:

[0133] ;

[0134] In the formula, Indicates time hour Current regional load Indicates time hour The volume of business within the current time window. express The maximum load that can be carried.

[0135] Then, the overall load level of the adaptive routing path is calculated, reflecting the congestion level of that path in the global network:

[0136] ;

[0137] In the formula, Indicates time Time communication node pair The overall load level of the adaptive routing path. Indicates the number of communication zones in the adaptive routing path. Indicates time hour Current regional load.

[0138] Constructing a congestion-aware penalty factor:

[0139] ;

[0140] In the formula, Indicates time Time communication node pair Congestion-aware penalty factor for adaptive routing paths Represents an exponential function. This represents the congestion sensitivity coefficient, used to adjust the strength of the load's suppression of feedback. Indicates time Time communication node pair The overall load level of the adaptive routing path.

[0141] The original immediate feedback value is adjusted by a congestion-aware penalty factor to obtain the immediate feedback value:

[0142] ;

[0143] In the formula, Indicates time Time communication node pair The immediate feedback value of the adaptive routing path. Indicates time Time communication node pair The original, real-time feedback value of the adaptive routing path. Indicates time Time communication node pair Congestion-aware penalty factor for adaptive routing paths.

[0144] In this step, when the overall load level is low, the congestion-aware penalty factor is close to 1, and the instantaneous feedback value basically maintains the original transmission result; when the overall load level increases, the congestion-aware penalty factor decreases rapidly, and even if the transmission is successful, its instantaneous feedback value will be significantly weakened.

[0145] Step S554: Update the state parameters using the real-time feedback value, and perform the next adaptive route selection using the updated state parameters.

[0146] In this step, the state parameters of the corresponding candidate arm are updated based on the penalty feedback and the selected adaptive routing path:

[0147] ;

[0148] ;

[0149] In the formula, and They represent candidate arms respectively. Deadline and The number of times the historical selection was made. and Representing time respectively and Candidate arm The estimated revenue, Indicates time Time communication node pair The immediate feedback value of the adaptive routing path.

[0150] In this step, the congestion-aware penalty mechanism in the feedback phase causes the revenue estimate of the corresponding candidate arm to decrease synchronously when a certain path is frequently selected by multiple communication nodes and the load increases. This causes some communication nodes to explore other candidate paths, thus achieving an adaptive adjustment effect when multiple communication nodes select concurrent routes.

[0151] Example 2:

[0152] like Figure 2 As shown, this embodiment provides a multi-layer network adaptive routing system based on stability constraints. The system includes:

[0153] The partitioning module is used to divide multiple wireless networks into communication areas to obtain a multi-layer topology representation that includes multiple communication areas, each of which includes multiple wireless network nodes.

[0154] The statistics module is used to collect historical transmission behavior data during the data forwarding process for multiple communication node pairs with communication needs in a multi-layer topology representation, and obtain the link historical state vector and the path end-to-end historical state vector.

[0155] The calculation module is used to calculate the comprehensive evaluation index of the link and the end-to-end success index of the path based on the historical state vector of the link and the historical state vector of the path end-to-end.

[0156] The filtering module is used to filter communication areas for stability based on comprehensive link evaluation indicators and construct a set of candidate area-level paths that meet stability constraints.

[0157] The routing module is used to perform adaptive routing selection for multi-armed gambling machines based on the candidate regional path set and the path end-to-end success rate index, so as to obtain the adaptive routing path for each communication node pair.

[0158] The partitioning module includes:

[0159] The first acquisition unit is used to acquire basic network information of multiple wireless networks, including identification information, communication relationships, link communication capability parameters, and location parameters of wireless network nodes.

[0160] The first building unit is used to construct a node-level network topology with wireless network nodes as vertices and communicable links as edges based on basic network information.

[0161] The partitioning unit is used to partition the wireless network nodes in the node-level network topology diagram into multiple communication regions according to the preset communication region partitioning rules.

[0162] The configuration unit is used to designate, for each communication area, wireless network nodes within that communication area that have communication connections with other communication areas as area interface nodes.

[0163] The second building unit is used to abstract each communication area as a region-level node, and to construct a multi-layer topology representation including node-level network topology and region-level network topology based on the communication connection relationship between region interface nodes.

[0164] The filtering module includes:

[0165] The second acquisition unit is used to acquire the comprehensive evaluation index of all communication links in each communication region in the multi-layer topology representation.

[0166] The first calculation unit is used to calculate the regional stability index of the communication area based on the comprehensive evaluation index of all links within the communication area.

[0167] The comparison unit is used to compare the regional stability index of the communication area with the preset stability threshold to determine whether the communication area meets the stability constraints.

[0168] The third acquisition unit is used to acquire multiple region-level transmission paths in the multi-layer topology representation, with the source region and target region corresponding to the communication node pair as the start and end regions;

[0169] The filtering unit is used to eliminate regional-level transmission paths that contain communication areas that do not meet stability constraints, thereby obtaining a set of candidate regional-level paths.

[0170] The routing module includes:

[0171] The fourth acquisition unit is used to acquire the corresponding candidate region-level path set for each communication node pair;

[0172] The third construction unit is used to construct multiple candidate arms based on the regional transmission paths in the candidate regional path set. The candidate arms include the access node and the exit node of each communication region in the regional transmission path.

[0173] The second calculation unit is used to calculate the historical revenue of the corresponding candidate arm by using the path end-to-end success rate index corresponding to multiple end-to-end transmission behaviors under the same regional transmission path.

[0174] An initialization unit is used to initialize the state parameters of the multi-armed gambling machine using the historical returns and historical selection counts of the candidate arms.

[0175] The selection unit is used for path selection strategies based on multi-armed gambling machines. It selects candidate arms based on state parameters to obtain adaptive routing paths for corresponding communication node pairs.

[0176] It should be noted that the specific methods by which each module performs operations in the system described in the above embodiments have been described in detail in the embodiments related to the method, and will not be elaborated here.

[0177] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0178] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An adaptive routing method for multilayer networks based on stability constraints, characterized in that, include: Multiple wireless networks are divided into communication areas to obtain a multi-layer topology representation comprising multiple communication areas. Each communication area includes multiple wireless network nodes, including: Acquire basic network information of multiple wireless networks, including identification information, communication relationships, link communication capability parameters, and location parameters of wireless network nodes; Based on basic network information, a node-level network topology is constructed with wireless network nodes as vertices and communicable links as edges; According to the preset communication area division rules, the wireless network nodes in the node-level network topology diagram are divided into multiple communication areas. For each communication zone, the wireless network node in that communication zone that has communication connections with other communication zones is designated as the zone interface node. Each communication region is abstracted as a region-level node, and a multi-layer topology representation including node-level network topology and region-level network topology is constructed based on the communication connection relationship between region interface nodes. For multiple communication node pairs with communication needs in a multi-layer topology representation, historical transmission behavior is statistically analyzed during data forwarding to obtain the link historical state vector and the path end-to-end historical state vector. Based on the link historical state vector and the path end-to-end historical state vector, calculate the link comprehensive evaluation index and the path end-to-end success index respectively. The stability of the communication area is screened by link comprehensive evaluation index, and a set of candidate area-level paths that meet the stability constraints is constructed. Based on the candidate regional path set and the path end-to-end success rate index, adaptive routing selection is performed for the multi-armed gambling machine to obtain the adaptive routing path for each communication node pair.

2. The adaptive routing method for multi-layer networks based on stability constraints according to claim 1, characterized in that, The process of obtaining the link historical state vector and the path end-to-end historical state vector includes: The historical transmission behavior information of all communication links in the multi-layer topology representation is statistically analyzed. The historical transmission behavior information includes the number of successful forwardings, the number of failures, the number of timeouts, the number of retransmissions, and the duration of continuous stable transmission. The historical state vector of each communication link is constructed based on the historical transmission behavior information of the communication link. The communication link is a link formed by a pair of wireless network nodes that can directly communicate wirelessly. In the multi-layer topology representation, communication node pairs with data communication needs are identified; For each communication node pair, obtain the actual end-to-end transmission path used during data forwarding, wherein the end-to-end transmission path includes multiple wireless network nodes passed through sequentially. By using the historical transmission behavior information of multiple end-to-end transmission paths of the same communication node pair, a path end-to-end historical state vector of the corresponding communication node pair is constructed.

3. The adaptive routing method for multi-layer networks based on stability constraints according to claim 1, characterized in that, The step of using comprehensive link evaluation indicators to screen communication areas for stability and constructing a set of candidate region-level paths that meet stability constraints includes: For each communication region in the multi-layer topology representation, obtain the comprehensive link evaluation index corresponding to all communication links within that communication region; Based on the comprehensive evaluation index of all links within the communication area, calculate the regional stability index of the communication area. The regional stability index of the communication area is compared with the preset stability threshold to determine whether the communication area meets the stability constraints. Using the source and target regions corresponding to the communication node pairs as the starting and ending regions, multiple region-level transmission paths are obtained in the multi-layer topology representation; By eliminating regional-level transmission paths that contain communication areas that do not meet stability constraints, a set of candidate regional-level paths is obtained.

4. The adaptive routing method for multi-layer networks based on stability constraints according to claim 1, characterized in that, The adaptive routing selection for the multi-armed gambling machine, based on the candidate regional path set and the end-to-end success rate index, yields an adaptive routing path for each communication node pair, including: For each pair of communication nodes, obtain the corresponding set of candidate region-level paths; Multiple candidate arms are constructed based on the regional transmission paths in the candidate regional path set. The candidate arms include the access node and the exit node of each communication region in the regional transmission path. The historical revenue of the corresponding candidate arm is calculated by using the path end-to-end success rate index corresponding to multiple end-to-end transmission behaviors under the same regional transmission path. Initialize the state parameters of the multi-armed gambling machine using the historical returns and historical selection counts of the candidate arms; Based on the path selection strategy of multi-armed gambling machines, candidate arms are selected according to state parameters to obtain the adaptive routing path of the corresponding communication node pair.

5. The adaptive routing method for multi-layer networks based on stability constraints according to claim 4, characterized in that, The path selection strategy based on the multi-armed gambling machine selects candidate arms according to state parameters to obtain an adaptive routing path for the corresponding communication node pair, including: Based on the path selection strategy of multi-armed gambling machines, candidate arms are selected according to state parameters to obtain regional-level routing paths; Determine the routing nodes for each communication region in the regional routing path to obtain the adaptive routing path for the corresponding communication node pair; After an end-to-end data transmission is completed on the adaptive routing path, a corresponding real-time feedback value is generated based on the actual transmission result. The state parameters are updated with real-time feedback values, and the next adaptive route selection is performed based on the updated state parameters.

6. A multi-layer network adaptive routing system based on stability constraints, characterized in that, include: The partitioning module is used to divide multiple wireless networks into communication areas to obtain a multi-layer topology representation that includes multiple communication areas, each of which includes multiple wireless network nodes. The statistics module is used to collect historical transmission behavior data during the data forwarding process for multiple communication node pairs with communication needs in a multi-layer topology representation, and obtain the link historical state vector and the path end-to-end historical state vector. The calculation module is used to calculate the comprehensive evaluation index of the link and the end-to-end success index of the path based on the historical state vector of the link and the historical state vector of the path end-to-end. The filtering module is used to filter communication areas for stability based on comprehensive link evaluation indicators and construct a set of candidate area-level paths that meet stability constraints. The routing module is used to perform adaptive routing selection for multi-armed gambling machines based on the candidate regional path set and the path end-to-end success index, so as to obtain the adaptive routing path for each communication node pair. The partitioning module includes: The first acquisition unit is used to acquire basic network information of multiple wireless networks, including identification information, communication relationships, link communication capability parameters, and location parameters of wireless network nodes. The first building unit is used to construct a node-level network topology with wireless network nodes as vertices and communicable links as edges based on basic network information. The partitioning unit is used to partition the wireless network nodes in the node-level network topology diagram into multiple communication regions according to the preset communication region partitioning rules. The configuration unit is used to designate, for each communication area, wireless network nodes within that communication area that have communication connections with other communication areas as area interface nodes. The second building unit is used to abstract each communication area as a region-level node, and to construct a multi-layer topology representation including node-level network topology and region-level network topology based on the communication connection relationship between region interface nodes.

7. The multi-layer network adaptive routing system based on stability constraints according to claim 6, characterized in that, The filtering module includes: The second acquisition unit is used to acquire the comprehensive evaluation index of all communication links in each communication region in the multi-layer topology representation. The first calculation unit is used to calculate the regional stability index of the communication area based on the comprehensive evaluation index of all links within the communication area. The comparison unit is used to compare the regional stability index of the communication area with the preset stability threshold to determine whether the communication area meets the stability constraints. The third acquisition unit is used to acquire multiple region-level transmission paths in the multi-layer topology representation, with the source region and target region corresponding to the communication node pair as the start and end regions; The filtering unit is used to eliminate regional-level transmission paths that contain communication areas that do not meet stability constraints, thereby obtaining a set of candidate regional-level paths.

8. The multi-layer network adaptive routing system based on stability constraints according to claim 6, characterized in that, The routing module includes: The fourth acquisition unit is used to acquire the corresponding candidate region-level path set for each communication node pair; The third construction unit is used to construct multiple candidate arms based on the regional transmission paths in the candidate regional path set. The candidate arms include the access node and the exit node of each communication region in the regional transmission path. The second calculation unit is used to calculate the historical revenue of the corresponding candidate arm by using the path end-to-end success rate index corresponding to multiple end-to-end transmission behaviors under the same regional transmission path. An initialization unit is used to initialize the state parameters of the multi-armed gambling machine using the historical returns and historical selection counts of the candidate arms. The selection unit is used for path selection strategies based on multi-armed gambling machines. It selects candidate arms based on state parameters to obtain adaptive routing paths for corresponding communication node pairs.

Images