A dynamic networking fast deployment method for portable repeaters

Abstract

The application discloses a kind of dynamic networking quick deployment methods for portable repeater, it is related to communication technical field, the method includes: step S1: in target repeater acquisition first heterogeneous parameter, monitoring initial relay parameter;Step S2: by monitoring route convergence time, abnormal node is screened out, and abnormal heterogeneous parameter and abnormal relay parameter are marked;Step S3: using abnormal heterogeneous parameter calculates first channel index, and then calculates correction relay parameter;Step S4: running correction relay parameter and acquisition second heterogeneous parameter, calculate second channel index and compare with first channel index, complete networking or return to execute step S3.The application is by acquisition heterogeneous parameter of heterogeneous service state, constructs channel index and accurate mapping model, realizes the continuous optimization congestion process of dynamic networking of self-convergent, with the advantages of improving data delivery success rate when dynamic networking and the beneficial effects of improving overall dynamic networking regulation efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of communication technology, in particular to a dynamic networking fast deployment method for a portable repeater. BACKGROUND

[0002] The portable repeater dynamic networking technology is a core means to realize regional fast and temporary communication coverage. The technology is to form a multi-hop mobile self-organizing network by multiple portable nodes autonomously, without relying on fixed infrastructure, with the core advantages of fast deployment, self-organization and self-healing. With the complication of application scenarios, the network needs to carry highly heterogeneous service streams, mainly including real-time audio streams, voice and key instructions, and network control signaling, etc. In the actual fast deployment process of using the portable repeater for dynamic networking, the coexistence of multiple heterogeneous data streams can easily cause serious protocol behavior conflicts, and then cause the sharp increase of the key network parameter of route convergence time, resulting in the significant decline of the network self-healing ability.

[0003] Most portable repeaters are equipped with only one or multiple independent radio frequency systems in pursuit of cost, size and power consumption, and all different types of service streams and control streams must compete on the same shared channel. In the process of using the portable repeater to co-transmit multiple heterogeneous data streams, the high-bandwidth audio stream and the TCP stream trying to maintain throughput will continuously occupy the wireless channel and data forwarding buffer of the node. This preemption is indiscriminate, which seriously squeezes the transmission opportunity required by the route control message, so that the topology-aware node will delay the discovery of link interruption, causing the reporting of topology change information to lag. Due to the serious delay of transmission and processing of control signaling, the response of the entire network to the topology change becomes extremely slow. Based on incomplete or outdated information, each node performs route calculation, resulting in that its routing table is in inconsistent state for a long time, and the route convergence time is deteriorated from the ideal state of several hundred milliseconds to several seconds or even tens of seconds. During the long convergence period, the communication channel will be blocked due to the congestion of multiple heterogeneous data streams, causing the control signaling data to be interrupted due to long-time transmission congestion, resulting in the interruption or loss of information instructions issued by the repeater, and further causing a large amount of delivery data to be unable to complete normal delivery transmission in the key time period at the initial stage of the dynamic networking process. In terms of data congestion caused by long route convergence time, the commonly used portable repeater lacks obvious and effective self-healing ability to relieve the channel data transmission, so that the portable repeater is difficult to achieve good data transmission effect when performing dynamic networking fast deployment for heterogeneous data streams. SUMMARY

[0004] The application provides a dynamic networking rapid deployment method for a portable repeater, and solves the problem that the existing portable repeater causes protocol behavior conflict due to multiple heterogeneous data streams in the process of deploying a dynamic network, so that the dynamic network channel is prone to congestion, and further causes poor network control signaling data transmission effect.

[0005] The application is achieved by the following technical solutions:

[0006] A dynamic networking rapid deployment method for a portable repeater, the method comprising:

[0007] Step S1: preset a reference time range for the route convergence time of a target repeater, preset an initial networking deployment for the target repeater to process heterogeneous data streams, monitor and collect first heterogeneous parameters composed of heterogeneous data streams in the initial networking deployment at all networking nodes of the target repeater, and monitor initial relay parameters of the target repeater for implementing the initial networking deployment;

[0008] Step S2: send control signaling of the initial networking deployment through the networking nodes, monitor the route convergence time, mark all networking nodes with route convergence time of the sent control signaling exceeding the reference time range as abnormal nodes, and mark the first heterogeneous parameters and the initial relay parameters collected by each abnormal node as abnormal heterogeneous parameters and abnormal relay parameters;

[0009] Step S3: for each abnormal node, calculate a first channel index representing the congestion amount of the networking channel using the abnormal heterogeneous parameters, construct a variable mapping model of the initial relay parameters and the first channel index, substitute the first channel index into the variable mapping model, and mark the calculation output result as a modified relay parameter representing the adjustment target of the abnormal relay parameter;

[0010] Step S4: update the modified relay parameter and send it to the repeater working point of the corresponding abnormal node, run the target repeater and collect second heterogeneous parameters, the collection method and data type of the second heterogeneous parameters are the same as those of the first heterogeneous parameters, calculate a second channel index using the second heterogeneous parameters, set a channel threshold value based on the first channel index, when the second channel index is less than the channel threshold value, it means that the channel resolution is effective, and when the second channel index is greater than the channel threshold value, return to execute step S3.

[0011] Further, the first heterogeneous parameters and the second heterogeneous parameters both include UDP stream bandwidth proportion values, TCP stream buffer proportion values and sending throughput proportion values obtained from each networking node;

[0012] The UDP flow bandwidth ratio value represents the ratio of the average rate of the UDP data flow to the physical bandwidth of the networking channel, the TCP flow buffer ratio value represents the ratio of the buffer occupied by the TCP protocol data to the total buffer of the networking node, and the sending throughput ratio value represents the ratio of the number of data bits sent to the maximum bit capacity of the networking channel.

[0013] Further, the UDP flow bandwidth ratio value is obtained by identifying the UDP flow of the target relay through a deep packet inspection method, the TCP flow buffer ratio value is obtained by querying the Socket buffer state of the dynamic networking kernel protocol stack, and the sending throughput ratio value is obtained by reading the sending queue state of the dynamic networking networking node.

[0014] Further, the initial relay parameter is set to the upper limit value of the MAC layer retransmission number of the target relay, and the upper limit value of the MAC layer retransmission number represents the maximum number of attempts for repeated sending of data frames in the target relay.

[0015] Further, the upper limit value of the MAC layer retransmission number is obtained by reading the object value of the management information base supporting the SNMP protocol; the reading process is completed by all networking nodes by initiating a request to the SNMP agent.

[0016] Further, a reference time range is set using a sliding window, and the process includes:

[0017] Initialize the sliding window to store the routing convergence time of the last several times, collect the actual time length value of each network convergence and add it to the tail of the sliding window, and remove the actual time length value data at the head when the number of actual time length value data in the sliding window exceeds the carrying capacity; calculate the mean and standard deviation of all actual time length values in the window, set a sensitivity coefficient, multiply the standard deviation by the sensitivity coefficient, and set the sum of the product and the mean as the upper limit value of the reference time range; when the routing convergence time exceeds the upper limit value, the networking node sending control signaling at this time is screened out and marked as an abnormal node.

[0018] Further, different sensitivity coefficients are independently assigned to different networking nodes in the dynamic network, wherein the networking nodes located in the core path of the network topology are set to the minimum value, and the edge networking nodes are set to the maximum value.

[0019] Further, the UDP flow bandwidth ratio value, the TCP flow buffer ratio value, and the sending throughput ratio value are respectively represented as U, Q, and T, the first weight α, the second weight β, and the third weight λ are set, the percentile function rank_pct is set, the channel penalty term ω is set, and the first channel index I1 is represented as I1,

[0020] The calculation formula of the first channel index I1 is represented as: ,

[0021] Here, a penalty threshold c is set, when When the channel penalty term ω=0, the rank_pct(U) in the formula represents the percentile risk ranking of the UDP stream bandwidth ratio among all network nodes, calculated by the percentile function.

[0022] Furthermore, the formula for calculating the channel penalty term ω is set as follows: ,

[0023] Where, ω max Indicates when At that time, the maximum feasible value of the channel penalty term ω.

[0024] Furthermore, within each network node, the product of the UDP flow bandwidth ratio and the TCP flow buffer ratio is obtained, and the penalty threshold c is set to the average of the product values ​​of all network nodes.

[0025] Furthermore, the variable mapping model is constructed based on the Sigmoid function, and its construction includes:

[0026] The maximum number of MAC retransmissions is set to R, which is the preset maximum number of MAC retransmissions in the target repeater. max Let the slope of the Sigmoid function be denoted as k, and let the center threshold of the Sigmoid function be preset and denoted as I. m Set the mapping penalty term φ.

[0027] The computational formula for the variable mapping model is then expressed as: ,

[0028] In the formula, R(I1) represents the upper limit of the number of MAC retransmissions corresponding to the first channel index in the variable mapping model.

[0029] Furthermore, in the calculation formula of the variable mapping model, the center weight value is multiplied by the center threshold to dynamically set the center threshold; the initial value of the center weight value is set to 1, and when two or more abnormal nodes appear consecutively in the initial network deployment, the center weight value is linearly increased; when two or more non-abnormal network nodes appear consecutively, the center weight value is linearly decreased.

[0030] Furthermore, critical link nodes and nodes with dense heterogeneous data flows are screened out from all network nodes; when critical link nodes and nodes with dense heterogeneous data flows are marked as abnormal nodes, the center weight value is adjusted first.

[0031] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0032] 1. By collecting heterogeneous parameters of heterogeneous service states, actively sensing the congestion impact of heterogeneous services on network channels, quantifying the service competition degree, and taking the route convergence timeout as a hard criterion to judge the network congestion risk in real time, the relay device can identify and control the transmission of control signaling in advance, and has self-flow control awareness;

[0033] 2. By constructing a channel index and an accurate mapping model, targeted parameter adjustment instead of blind adjustment of the entire network is realized. The corrected relay parameters obtained by mapping are used for abnormal nodes, and the blocked link corresponding nodes and parameters are automatically labeled. The network performance deterioration is changed from global ambiguity to local controllability, which improves the regulation efficiency and accuracy of the overall dynamic networking;

[0034] 3. The automatic judgment of the effect of relieving is set, the closed-loop verification mechanism is constructed to improve the network self-healing speed, the control signaling is restored to priority transmission, and the continuous optimization process of self-adjustment and self-convergence is formed. The data delivery success rate and network connectivity reliability in the initial stage of deployment are significantly improved, and the communication continuity and recovery ability of the portable relay device in real-time tasks are guaranteed. BRIEF DESCRIPTION OF DRAWINGS

[0035] The drawings described herein are used to provide further understanding of the embodiments of the present application, constitute a part of the present application, and do not constitute a limitation on the embodiments of the present application. In the drawings:

[0036] Figure 1 The structure diagram of the present application. DETAILED DESCRIPTION

[0037] In order to make the purpose, technical scheme and advantages of the present application more clear and obvious, the present application is further described in detail below in combination with embodiments and drawings. The illustrative embodiments of the present application and their descriptions are only used to explain the present application, and do not constitute a limitation on the present application.

[0038] Embodiment 1, as shown in the figure, the present embodiment is a dynamic networking fast deployment method for portable relay device, the method comprises: Figure 1

[0039] Step S1: preset a reference time range for the route convergence time of the target relay device, preset the initial networking deployment for the target relay device to process heterogeneous data flow, monitor and collect the first heterogeneous parameter composed of heterogeneous data flow in the initial networking deployment of the target relay device, and monitor the initial relay parameter used by the target relay device to implement the initial networking deployment;

[0040] ​Step S2: sending control signaling of initial networking deployment through networking nodes, monitoring route convergence time, marking all networking nodes with route convergence time of sending control signaling exceeding reference time range as abnormal nodes, and marking first heterogeneous parameters and initial relay parameters collected by each abnormal node as abnormal heterogeneous parameters and abnormal relay parameters correspondingly;

[0041] Step S3: for each abnormal node, using abnormal heterogeneous parameters to calculate a first channel index representing channel congestion amount, constructing a variable mapping model of initial relay parameters and first channel index, substituting first channel index into variable mapping model, and marking calculation output result as a modified relay parameter representing adjustment target of abnormal relay parameter;

[0042] Step S4: updating modified relay parameter and sending to relay working point of corresponding abnormal node, running target relay and collecting second heterogeneous parameters, the collection method and data type of second heterogeneous parameters are the same as those of first heterogeneous parameters, calculating second channel index using second heterogeneous parameters, setting channel threshold based on first channel index, when second channel index is less than channel threshold, it means that channel resolution takes effect, and when second channel index is greater than channel threshold, return to execute step S3.

[0043] The quick deployment process of dynamic networking using portable repeaters can be summarized as the following steps: power-on self-test start, scanning neighbor nodes and establishing links, generating multi-hop topology network, dynamic convergence of routing protocol and data communication data distribution. The reference time range represents an expected normal range of routing convergence time based on the processing capacity of the target repeater, the protocol mechanism and the typical topology size. As long as the actual convergence time falls within the reference time range, the control signaling transmission and topology update are considered to be in a healthy state. The initial networking deployment means that before the target repeater starts dynamic networking, in order to ensure that it can simultaneously carry multiple different types of data services, the initial network working configuration is set for all networking nodes, so that they can have the ability to perform basic forwarding, competition and scheduling of heterogeneous data streams in the initial stage. In specific implementation, the initial networking deployment can include default configuration related to radio frequency and physical layer, initial setting of MAC layer resource competition and queuing strategy, setting of initial parameters of routing protocol and initial threshold of traffic management, etc., which can include, for example, initial setting of parameters such as reference channel, bandwidth, transmission power, default priority of each service flow, competition window size, configuration of topology discovery and update process start, and default bandwidth constraint for high-bandwidth audio stream and TCP stream. The all networking nodes refer to the collection of all node entities that form network connection through wireless link and participate in dynamic networking topology maintenance in the initial networking deployment. The first heterogeneous parameter represents various types of heterogeneous service flow data in the initial networking deployment process, which can be a data or a group of data in specific application; the specific parameter can be distinguished by service type source, such as voice flow parameter, real-time audio flow parameter, TCP data flow parameter, control signaling state parameter, etc. The initial repeater parameter represents the basic operating parameters set by the device for participating in the network when the portable repeater starts to perform dynamic networking deployment, which includes MAC layer window parameter, physical layer channel parameter, network layer control parameter and other parameter categories in actual application, which can be further set as data frame upload times, CTS trigger threshold, node transmission power, maximum data frame length, relay buffer space capacity, etc.; in specific application, it can be a data or a group of data like the first heterogeneous parameter.

[0044] The core target of screening abnormal nodes by monitoring the route convergence time is to identify which networking nodes cause channel congestion and lead to slow route convergence in the initial stage of networking by controlling the transmission performance of signaling. In the initial networking deployment stage, the target relay will let all the networking nodes participating in the deployment send control signaling in the routing protocol; monitor the time taken for the entire network to complete topology convergence, that is, the time experienced from the start of route initialization to the consistency of the routing table of each networking node. If this period is extended, it means that the control signaling is blocked or delayed. If the control signaling sent by a certain networking node causes the convergence time to exceed the reference time range, it means that the link or forwarding behavior related to the node has a risk of congestion; the parameters corresponding to these networking nodes will be considered to indicate the location of congestion, which will be used for subsequent calculation and relay parameter modification. The abnormal heterogeneous parameters and abnormal relay parameters are special first heterogeneous parameters and initial relay parameters. The first channel index is a single numerical value calculated from the abnormal heterogeneous parameters, which is used to quantify the congestion degree of the network channel where the abnormal node is located; it is used to compress the multi-dimensional heterogeneous parameters into a scalar index, which is convenient for mapping to the subsequent relay parameter adjustment model. The meaning of steps S2-S3 is that all abnormal nodes are screened out by route convergence time and the corresponding abnormal heterogeneous parameters are collected, for each abnormal node, the first channel index representing the congestion degree of the channel is calculated using the abnormal heterogeneous parameters, and the relay parameter required to maintain channel balance is calculated using the quantized value, and this relay parameter is represented as a modified relay parameter. The data type of the modified relay parameter is consistent with the initial relay parameter, which is a dynamic adjustment value calculated for a specific abnormal node and a specific congestion state, and will change according to the congestion state.

[0045] The modified relay parameter is updated to the working point of the relay corresponding to the abnormal node, the abnormal node currently performing data calculation changes the deployment behavior, and attempts to relieve the congestion caused by heterogeneous data flow; after updating the parameter, the target relay continues to run for a period of time, and the heterogeneous parameter is re-collected using the same index as the initial collection of the first heterogeneous parameter, and the heterogeneous parameter collected after running based on the modified relay parameter is marked as the second heterogeneous parameter, that is, the collection and calculation process of the first heterogeneous parameter is the same as that of the second heterogeneous parameter. Similarly, at this time, the first channel index calculation method is used again, and the second heterogeneous parameter is used to calculate a new channel index, which is marked as the second channel index, that is, the properties of the second channel index are the same as those of the first channel index, and the difference lies in the numerical size. The numerical size of the channel threshold is smaller than that of the first channel index, and the channel threshold is used to serve as a verification standard for whether the second channel index meets the current networking channel congestion relief. The channel threshold is set based on the first channel index, which is used to reduce the influence of the high first channel index, indicating that even if the channel is not completely congested, there is still high congestion. The channel threshold is directly compared with the second channel index, and since the second channel index represents that the higher the numerical value, the more congested the channel, when the second channel index is less than the channel threshold, it indicates that the adjusted relay parameter has successfully reduced the network congestion, and the abnormal node does not need to be further adjusted, the system can continue to operate normally, the routing convergence performance is improved, and the control signaling transmission is restored. When the second channel index is greater than the channel threshold, it indicates that the modified relay parameter does not have the expected effect, and the channel congestion is even more severe, or the adjustment is not enough to relieve the congestion. At this time, the embodiment selects to re-calculate the modified relay parameter, and in specific application, different degrees of adjustment range or calculation adjustment strategy can be adjusted for the calculation process, so as to again attempt to relieve the congestion and realize closed-loop feedback. In this embodiment, the system does not rely only on one parameter adjustment, but realizes self-adaptive optimization of the abnormal node through the cycle of "measurement-judgment-improvement-measurement". In actual engineering, threshold, minimum step or multi-window average can be combined to avoid over-adjustment or oscillation phenomenon.

[0046] In specific implementation, the setting method of the channel threshold can not be uniquely limited. For example, the congestion degree of the current networking channel can be evaluated according to the routing convergence time length when the networking node channel is congested in the existing relay technology, combined with the routing convergence time of the current target networking node, the theoretical maximum value of the first channel index is estimated in the form of equal percentage based on the experience rule, and the channel threshold is set according to the percentage (such as 50%-60% numerical value) based on the theoretical maximum value. The numerical size of the first channel index can also be directly selected to set the channel threshold.

[0047] In a specific implementation, the first channel index can be calculated by a normalization method or an empirical threshold quantization method on the abnormal heterogeneous parameters, for example, setting high, medium, and low levels for each heterogeneous parameter, and accumulating the level scores of each parameter according to different contribution degrees to form the final first channel index. As a specific application, the process can be set as follows: determining the index set of the heterogeneous parameters and the sampling window, mapping each index to an index score of different discrete levels after using linear normalization to set each index to a unified scale, assigning a contribution weight to each index score, calculating the weighted score and adding a penalty term to obtain the first channel index. The linear normalization method can use a clip function normalization method, and the calculation formula can be set as follows: where x represents the original parameter value of each heterogeneous parameter, x min and x max represent the sample minimum value and the sample maximum value, respectively. The clip function represents the calculation result in the fraction ; if the calculation result of the fraction is less than 0, x takes 0, if it is greater than 1, x takes 1, and the rest of the results keep the original value. This is used to prevent extreme values or sampling noise from causing the normalization result to exceed the (0, 1) interval and prevent abnormal values from exceeding the range.

[0048] The variable mapping model represents the mapping of the current initial relay parameter of the repeater to the network channel congestion represented by the first channel parameter, and establishes a calculation function relationship. In a specific implementation, the variable mapping model can be an empirical simulation formula, a rule mapping table, a regression calculation model, or a nonlinear function. As a specific application, the variable mapping model can be constructed using a simulation formula, and the process can be set as follows: setting the first channel index as I1, the initial relay parameter as R0, and the modified relay parameter as R, and constructing a segmented simulation calculation formula: where k1 and k2 represent the first fitting coefficient and the second fitting coefficient, respectively, I t represents the fitting preset threshold, and R(I1) represents the modified relay parameter corresponding to the first channel index I1. The fitting preset threshold represents the quantization reference value of the higher point of the channel congestion degree. When the first channel index is less than the fitting preset threshold, the network is relatively smooth, and the system adopts a linear growth rule to adjust the upper limit of retransmission; when the first channel index is greater than the fitting preset threshold, the network has obvious congestion, and the growth rule changes to a logarithmic type, so that the retransmission number adjustment tends to be moderate or saturated, preventing the system from over-retransmitting in the high congestion interval.

[0049] In embodiment 2, the reference time range is set using a sliding window, and the process includes:

[0050] The sliding window is initialized to store the route convergence time of the last several times, the actual time length value of each network convergence is collected and added to the tail of the sliding window, and when the actual time length value data in the sliding window exceeds the carrying capacity, the actual time length value data at the head is removed; the time length mean value of all the actual time length values in the window and the time length standard deviation are calculated, a sensitivity coefficient is set, the time length standard deviation is multiplied by the sensitivity coefficient, and the sum of the product and the time length mean value is set as the upper boundary value of the reference time length range; when the route convergence time exceeds the upper boundary value, it is determined that the networking node sending the control signaling at this time is screened out and marked as an abnormal node; the sensitivity coefficient is independently allocated to different networking nodes in the dynamic networking, wherein the networking node located on the core path of the network topology is set to the minimum value, and the edge networking node is set to the maximum value.

[0051] The initialization of the sliding window process represents the creation of a fixed-capacity data structure for storing the network route convergence time of the last several times. Each time the network completes route convergence, the actual time value monitored is added to the tail of the sliding window; since the sliding window has a fixed carrying capacity, when new data is added, the oldest data is deleted, ensuring that the statistical results are based only on the recent state. The mean value in the sliding window represents the average level of the network convergence time of the last several times, representing the central value of the time required for normal network convergence, providing a reference value. The standard deviation in the sliding window represents the fluctuation range of the route convergence time, and the greater the fluctuation, the more likely there is a short congestion or network instability, providing a sensitivity reference value. The sensitivity coefficient is used to adjust the tolerance of the system to fluctuations, and when the sensitivity coefficient is higher, the upper boundary value of the reference time length range is higher, and the node is less likely to be judged as abnormal; when the sensitivity coefficient is lower, the upper boundary value of the reference time length range is lower, and the node is more likely to be judged as abnormal. In specific implementation, the sensitivity coefficient of the networking nodes other than the core path and the edge point in the dynamic networking can be allocated using a segmented constant allocation method, and for the networking topology, the path distance index of each node from the core path is calculated and the segmented intervals are set, and for the networking nodes in each segmented interval, the fixed sensitivity coefficient is set along the downward trend of the value from the core path to the edge point. As a specific implementation, the calculation form of the upper boundary value can be represented as: the upper boundary value is represented as ε, the mean value and the standard deviation of the route convergence time are represented as γ and ρ respectively, and the sensitivity coefficient is represented as v, then the upper boundary value calculation form is represented as wherein represents the tolerance of the fluctuation adjusted according to the sensitivity coefficient. It should be noted that since the time length data is a scalar that only measures one-sided value, the boundaries of the reference time length range can usually not be set to the lower boundary, and in specific application, the lower boundary value of the reference time length range can be set to 0 or a small value.

[0052] In this embodiment, the first heterogeneous parameter and the second heterogeneous parameter each include a UDP flow bandwidth proportion value, a TCP flow buffer proportion value, and a sending throughput proportion value obtained from each networking node;

[0053] The UDP flow bandwidth proportion value represents a ratio of an average rate of a UDP data stream to a physical bandwidth of a networking channel, the TCP flow buffer proportion value represents a ratio of a buffer occupied by a TCP protocol data to a total buffer of a networking node, and the sending throughput proportion value represents a ratio of a number of data bits sent to a maximum bit capacity of the networking channel. The UDP flow bandwidth proportion value is obtained by identifying a UDP flow of a target repeater through a deep packet inspection method, the TCP flow buffer proportion value is obtained by querying a Socket buffer state of a kernel protocol stack of a dynamic networking, and the sending throughput proportion value is obtained by reading a sending queue state of the dynamic networking.

[0054] The UDP flow bandwidth proportion value is equal to a ratio of an average rate of a UDP data stream to a physical bandwidth of a networking channel. The UDP flow bandwidth proportion value is selected because UDP is usually used for real-time audio, video, or instruction streams, and is characterized by high bandwidth, sensitivity to delay, and no congestion control, and can accurately reflect a proportion of a channel occupied by a UDP flow. In calculation, a proportion of resources occupied by the UDP flow in the entire network channel can be quantified, so as to measure potential competitive pressure of the UDP flow on other data flows and control signaling. In specific application, DPI can be used to identify average rate data of a UDP flow of a target repeater.

[0055] The TCP flow buffer proportion value represents a ratio of a buffer occupied by TCP data to a total buffer of an abnormal node. TCP protocol adjusts a data sending rate through a sliding window and congestion control, and data is occupied by a Socket buffer of a node before being sent. The Socket buffer is a network communication endpoint provided by a system, and is used for data transmission between processes or devices. In a TCP protocol stack, the Socket buffer is a memory area used for temporarily storing data. The higher the TCP flow buffer proportion value in an abnormal node is, the more the forwarding capability of the abnormal node is compressed, which causes control signaling to be delayed or blocked. In a portable repeater, multiple flows share a single channel, and TCP flow is easy to continuously occupy a buffer due to high throughput, which affects forwarding of low-quantity control packets. The TCP flow buffer proportion value is included in the index, so that the first channel index can reflect a more real congestion situation. In specific implementation, the TCP flow buffer proportion value can be obtained by embedded flow monitoring or querying a Socket buffer state of a kernel protocol stack of a dynamic networking through a Netlink interface.

[0056] The sending throughput ratio value provides an overall load index for assisting calculation of the UDP flow bandwidth ratio value and the TCP flow buffer ratio value by a ratio of a number of bits actually sent by the node to a maximum capacity of the physical channel, and can complementarily reflect a stress level of the node on the entire network channel. When the sending throughput approaches an upper limit of the channel, it indicates that the node can have congestion problems such as queue backlog and blocked control signaling. By quantifying the index, the channel index calculation can be directly triggered to determine whether the relay parameters need to be optimized. By monitoring the overall load of the node, a high-load node can be found in time, a single flow index failure can be avoided, and the transmission opportunity of control messages can be preferentially protected to reduce the route convergence delay.

[0057] Further, as a feasible implementation, the initial relay parameter is set as a MAC layer retransmission upper limit value of the target relay, the MAC layer retransmission upper limit value representing a maximum number of attempts of repeated sending of a data frame in the target relay; the MAC layer retransmission upper limit value is obtained by reading an object value of a management information base supporting an SNMP protocol; and the reading process is completed by all the networking nodes by initiating a request to an SNMP agent.

[0058] The MAC layer retransmission mechanism is used for repeated sending of a data frame when there is packet loss or collision on a wireless link. The MAC layer retransmission upper limit value is selected to determine the maximum number of attempts of the node when there is a short-time transmission failure, thereby guaranteeing the transmission success rate of critical control signaling and data frames; the initial retransmission upper limit value ensures that critical route messages and control instructions can still be successfully transmitted under the interference of multiple heterogeneous flows, thereby shortening the route convergence time. A reasonable upper limit prevents the node from occupying the channel due to excessive retransmission, and provides transmission opportunities for UDP and TCP heterogeneous data flows. When the MAC layer retransmission number of the networking node is too low, the packet loss rate of the networking channel is high, which affects the route convergence and the reliability of control signaling; when the MAC layer retransmission number of the networking node is too high, the data frame occupies more time in the channel, which can cause congestion of heterogeneous service flows. When setting the initial relay parameter, a conventional value according to an empirical rule can be set as an initial reference value. The SNMP is a lightweight, universal and programmable interface, and the MAC layer parameters of each node can be remotely obtained without modifying the underlying protocol stack. The process of reading the management information base through the SNMP is that the networking node initiates a request to the SNMP agent of the target relay, reads the object value of the MAC retransmission upper limit value in the MIB, and obtains the result as the initial relay parameter for dynamic networking deployment.

[0059] Further, as a feasible implementation, the UDP flow bandwidth ratio value, the TCP flow buffer ratio value and the sending throughput ratio value are respectively represented as U, Q and T, a first weight α, a second weight β and a third weight λ are set, a percentile function rank_pct is set, a channel penalty term ω is set, and a first channel index is represented as I1,

[0060] The calculation formula of the first channel index I1 is represented as: ,

[0061] wherein a penalty threshold c is set, when the channel penalty term ω=0; rank_pct(U) in the formula represents the risk ranking percentile of the UDP flow bandwidth proportion value in all networking nodes calculated by the percentile function.

[0062] The represents simultaneously considering two main loads of the UDP flow bandwidth proportion value and the TCP flow buffer proportion value in the abnormal node; when the value is high, it indicates that the node simultaneously has high UDP flow bandwidth proportion value and TCP flow buffer proportion value, and the abnormal node may be congested seriously; when the value is low, it indicates that the current abnormal node is under light load or single flow occupation, and the channel penalty term is not triggered when it is lower than the penalty threshold.

[0063] The first weight α is used to adjust the contribution of the UDP flow bandwidth proportion value to the channel index, the second weight β is used to adjust the contribution of the TCP flow buffer proportion value to the channel index, and the third weight λ is used to adjust the contribution of the sending throughput proportion value (i.e. the overall sending load of the node) to the channel index. The channel penalty term ω represents additional adjustment weighting of the channel index, which is used to strengthen the risk index of the high-load node and ensure that the congested node is identified in priority in the index. The relationship between the UDP flow bandwidth proportion value and the channel congestion varies due to physical layer performance and service scheduling strategy, and it is difficult to make unified risk judgment in a heterogeneous device and a multi-bandwidth scene using the absolute proportion, so the percentile function rank_pct is used to represent the relative congestion position in the embodiment. Because UDP is usually real-time service priority, its preemption is rigid, short cycle, and is the most direct source of pressure on the route, so in the specific implementation, the first weight α can be set as the maximum weight proportion. The TCP flow buffer proportion value Q is set in the form of a square term because the risk of the TCP flow buffer proportion value accelerates the deterioration of the networking channel congestion condition as the value increases, which is a strong nonlinear parameter, and the channel occupation risk is accelerated; setting the square term can increase the data sensitivity and increase the presentation of the risk difference value.

[0064] Further, based on the calculation formula of the first channel index I1 and each parameter in the formula, in the embodiment, as a feasible specific application, the channel threshold can be further limited, and the setting process of the channel threshold includes: setting the channel threshold as Th, setting the reduction coefficient η greater than 1,

[0065] The channel threshold Th calculation formula of the first channel index I1 is represented as: .

[0066] The reduction coefficient η represents the adjustment compression strength of the control channel threshold value relative to the first channel index I1. The greater the adjustment of the reduction coefficient η, the smaller the channel threshold value, and the more aggressive the identification of high congestion channels, that is, the greater the improvement required to determine the successful decongestion. In the calculation formula, when the first channel index I1 is smaller, it indicates that the channel is relatively smooth at this time, and the channel threshold value should not limit the second channel index, so the channel threshold value Th tends to the first channel index I1 at this time. When the first channel index I1 is larger, the channel threshold value Th will also be larger, but it is affected by the root, and the channel threshold value Th tends to sublinear growth, and the growth rate slows down significantly. The greater the first channel index I1, the closer the current networking channel may be to a completely congested state, and in order to avoid the existence of transitional screening, the channel threshold value cannot be set too low, so the channel threshold value is still positively correlated with the first channel index, but a significant networking channel redundancy space needs to be reserved, so that the difference between the channel threshold value and the first channel index cannot be too low, and the root is processed in the denominator.

[0067] In specific application, the form of the percentile function rank_pct can be set by sorting method, discrete distribution method, piecewise mapping, etc. As a specific implementation, if the discrete percentile method is used, there are N abnormal nodes, and the UDP flow bandwidth proportion value is represented as U1, U2,..., UN according to the abnormal node serial number. All nodes are arranged in ascending order of UDP flow bandwidth proportion value, and the order of the UDP flow bandwidth proportion value of all abnormal nodes after sorting is: U(1)<U(2)<…<U(N); Assuming that abnormal node j is arranged in the fth position after sorting, the value of the percentile function of abnormal node j is represented as: rank_pct(Uj)=(f-1) / (N-1).

[0068] Further, as a feasible implementation, the calculation formula of the channel penalty term ω is set as:

[0069] ,

[0070] wherein ω max represents the maximum implementable value of the channel penalty term ω when The product value of the UDP flow bandwidth proportion value and the TCP flow buffer proportion value is obtained in each networking node, and the value of the penalty threshold c is set to the average value of the product values of all networking nodes.

[0071] The ω max represents the maximum theoretical value of the actual implementable channel penalty term, which can be set according to empirical rules, historical or real-time data statistics. In the formula, ω Part represents the current abnormal node load exceeding the penalty threshold value, measures the severity of exceeding the threshold value. (1-c) is used to ensure that the penalty term does not exceed the set upper limit, keeping the channel index stable; when Approaching 1 makes the value of ω approach ω max . The value of the penalty threshold c is set to the product value mean of all networking nodes , which means that c is set by the average product of all network nodes, indicating that the penalty threshold is adaptively adjusted with the network load. When the overall load is high, the penalty threshold c becomes high, so that the mechanism only punishes relatively higher abnormal nodes; when the overall load is low, the penalty threshold c becomes low, so that the network appears to be slightly congested and may trigger the penalty. Using the product instead of a single value emphasizes the risk of double load superposition and avoids false positives caused by a single flow occupying too high. The product value mean is used as the threshold value to prevent slight fluctuations from triggering ω, improving the stability of the channel index calculation. The product value mean is used as the threshold value to prevent slight fluctuations from triggering ω, improving the stability of the channel index calculation.

[0072] Further, as a feasible implementation, the variable mapping model is constructed based on a Sigmoid function, and the construction content includes:

[0073] Setting the upper limit value of MAC retransmission number as R, which is preset as the maximum value of MAC retransmission number R max in the target relay, setting the Sigmoid slope as k, and presetting the center threshold value of the Sigmoid function as I m , setting the mapping penalty term φ,

[0074] The calculation formula of the variable mapping model is: ,

[0075] Wherein, R(I1) in the formula represents the upper limit value of the MAC retransmission number corresponding to the first channel index in the variable mapping model.

[0076] The center threshold value represents the intermediate risk level of the abnormal node load, which can be set according to the empirical rule. When the first channel index is less than the center threshold value, the Sigmoid output is close to 1, and the value of R is close to the maximum value R max , at this time the risk is low, and the retransmission number can be increased to ensure the data transmission reliability of the networking channel; when the first channel index is greater than the center threshold value, the upper limit value of the MAC retransmission number R decreases significantly, at this time the risk is high, and the retransmission number needs to be reduced to relieve the channel pressure. The Sigmoid slope represents the sensitivity of the upper limit value of the MAC retransmission number to the change of the first channel index. The mapping penalty term is used to compensate for the adjustment, to ensure that the mapped retransmission number R will not be too high due to the too low first channel index. The maximum value of the MAC retransmission number R maxThe maximum allowed value of the number of MAC retransmissions in dynamic networking is represented, which is used to define the upper limit of the mapping, to ensure the reliability of low-load nodes, so that abnormal nodes can still fully retry data in the case of low channel index, and to ensure the success of control signaling transmission. The first channel index is mapped to the upper limit value of the number of MAC retransmissions by the Sigmoid function, realizing nonlinear adaptive adjustment of the node load. Low-load nodes maintain a high number of retransmissions to ensure data reliability, and high-load nodes reduce the number of retransmissions to alleviate channel congestion. The center threshold and slope of the Sigmoid and the mapping penalty term provide a controllable adjustment space, thereby supporting fast deployment and closed-loop self-healing optimization of dynamic networking.

[0077] Further, as a feasible implementation, in the calculation formula of the variable mapping model, the center weight value is multiplied by the center threshold value to dynamically set the center threshold value; the initial value of the center weight value is 1, when two or more abnormal nodes appear continuously in the initial networking deployment, the center weight value is linearly increased; when two or more non-abnormal networking nodes appear continuously, the center weight value is linearly decreased; the key link nodes and the heterogeneous data flow intensive nodes are screened out among all the networking nodes; when the key link nodes and the heterogeneous data flow intensive nodes are labeled as abnormal nodes, the center weight value is adjusted preferentially.

[0078] The center threshold value of the Sigmoid mapping model is no longer fixed, but is dynamically adjusted according to the network running state: the sensitivity is automatically increased in high load, and the sensitivity is reduced in low load. At the same time, the key link and the heterogeneous flow intensive node are preferentially considered to ensure the smoothness of the core communication link and reduce the overall routing convergence delay and data blocking risk. The key link nodes and the heterogeneous data flow intensive nodes are screened out in the network, and when these nodes are labeled as abnormal nodes, they are preferentially adjusted to ensure that the congestion of the core path and the traffic intensive node is preferentially relieved. Through the dynamic adjustment of the continuously appearing abnormal and non-abnormal node information, adaptive adjustment of network congestion identification is realized, avoiding the misjudgment or slow reaction problem caused by the fixed threshold value. The center weight value is multiplied by the center threshold value, so that the inflection point of the Sigmoid function is dynamically adjusted, and the key link and the heterogeneous data flow intensive node are preferentially considered through the linear increase and decrease of the continuous abnormal or non-abnormal node information, realizing the closed-loop optimization of adaptive adjustment of the number of MAC retransmissions, improving the abnormal node identification accuracy and the network self-healing ability, and ensuring the fast deployment and stable operation of dynamic networking in a multi-heterogeneous flow environment.

[0079] The above specific embodiments further illustrate the purpose, technical solutions and advantages of the present application. It should be understood that the above description is only a specific embodiment of the present application and is not used to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for dynamic networking and rapid deployment of portable repeaters, characterized in that, The method comprises: Step S1: presetting a reference time range for the route convergence time of the target relay, presetting an initial networking deployment for the target relay to process heterogeneous data streams, monitoring and collecting first heterogeneous parameters composed of the heterogeneous data streams in the initial networking deployment at all networking nodes of the target relay, and monitoring initial relay parameters of the target relay for implementing the initial networking deployment; Step S2: sending control signaling of the initial networking deployment through the networking nodes, monitoring the route convergence time, marking all networking nodes with a route convergence time of the sent control signaling exceeding the reference time range as abnormal nodes, and marking the first heterogeneous parameters and the initial relay parameters collected by each abnormal node as abnormal heterogeneous parameters and abnormal relay parameters; Step S3: for each abnormal node, calculating a first channel index representing the congestion amount of the networking channel by using the abnormal heterogeneous parameters, constructing a variable mapping model of the initial relay parameters and the first channel index, substituting the first channel index into the variable mapping model, and marking the calculation output result as a modified relay parameter representing an adjustment target of the abnormal relay parameter; Step S4: updating the modified relay parameter and sending it to the relay working point of the corresponding abnormal node, running the target relay and collecting second heterogeneous parameters, the collection method and data type of the second heterogeneous parameters being the same as those of the first heterogeneous parameters, calculating a second channel index by using the second heterogeneous parameters, setting a channel threshold value based on the first channel index, and when the second channel index is less than the channel threshold value, indicating that the channel resolution is effective, and when the second channel index is greater than the channel threshold value, returning to execute step S3; The first heterogeneous parameters and the second heterogeneous parameters both comprise UDP stream bandwidth proportion values, TCP stream buffer proportion values and sending throughput proportion values obtained from each networking node; The UDP stream bandwidth proportion value represents the ratio of the average rate of the UDP data stream to the physical bandwidth of the networking channel, the TCP stream buffer proportion value represents the ratio of the buffer occupied by the TCP protocol data to the total buffer of the networking node, and the sending throughput proportion value represents the ratio of the number of data bits sent to the maximum bit capacity of the networking channel; The initial relay parameter is set as an upper limit value of the MAC layer retransmission number of the target relay, and the upper limit value of the MAC layer retransmission number represents the maximum number of attempts for repeated sending of data frames in the target relay; The UDP stream bandwidth proportion value, the TCP stream buffer proportion value and the sending throughput proportion value are respectively represented as U, Q and T, a first weight α, a second weight β and a third weight λ are set, a percentile function rank_pct is set, a channel penalty term ω is set, and the first channel index is represented as I1, The calculation formula of the first channel index I1 is represented as: , wherein a penalty threshold c is set, and when the channel penalty term ω = 0; rank_pct(U) in the formula represents a risk ranking percentile of the UDP flow bandwidth proportion value in all networking nodes calculated by a percentile function. The variable mapping model is constructed based on a Sigmoid function, and the construction content comprises: The upper limit of the number of MAC retransmissions is denoted as R, and the maximum number of MAC retransmissions preset in the target repeater is R max The center threshold of the Sigmoid function is denoted as I m The mapping penalty term φ is set, The computational formula of the variable mapping model is represented as: , In the formula, R(I1) represents the upper limit value of the MAC retransmission number corresponding to the first channel index in the variable mapping model; The setting process of the channel threshold value comprises: setting the channel threshold value as Th, setting a reduction coefficient η greater than 1, The channel threshold Th of the first channel index I1 is expressed as: .

2. The method of claim 1, wherein, The UDP flow bandwidth proportion value is obtained by identifying the UDP flow of the target relay through a deep packet inspection method, the TCP flow buffer proportion value is obtained by querying the state of the Socket buffer of the dynamic networking kernel protocol stack, and the sending throughput proportion value is obtained by reading the state of the sending queue of the dynamic networking node.

3. The method of claim 1, wherein, The upper limit of the MAC layer retransmission times is obtained by reading the object value of the management information base supporting the SNMP protocol, and the reading process is completed by all the networking nodes by initiating a request to the SNMP agent.

4. The method of claim 1, wherein, The reference duration range is set using a sliding window, and the process includes: The sliding window is initialized to store the convergence time of the latest several routes, the actual duration value of each network convergence is collected and added to the tail of the sliding window, and when the number of actual duration value data in the sliding window exceeds the carrying capacity, the actual duration value data at the head is removed; the duration mean value of all actual duration values in the window is calculated, a sensitivity coefficient is set, the product of the duration standard deviation and the sensitivity coefficient is multiplied, and the sum of the product and the duration mean value is set as the upper boundary value of the reference duration range; when the route convergence time exceeds the upper boundary value, the networking node sending control signaling at this time is screened out and marked as an abnormal node.

5. The method of claim 4, wherein, Different sensitivity coefficients are independently allocated to different networking nodes in the dynamic network, wherein the networking nodes located in the core path of the network topology are set to the minimum value, and the edge networking nodes are set to the maximum value.

6. The method of claim 1, wherein, The computation of the channel penalty term ω is set as: , where ω max represents the maximum implementable value of the channel penalty term ω when the maximum implementable value of the channel penalty term ω.

7. The method of claim 6, wherein, The product value of the UDP flow bandwidth proportion value and the TCP flow buffer proportion value is obtained in each networking node, and the value of the penalty threshold c is set to the mean value of the product values of all networking nodes.

8. The method of claim 1, wherein, In the calculation formula of the variable mapping model, the center weight value is multiplied by the center threshold value to dynamically set the center threshold value; the initial value of the center weight value is 1, and when two or more abnormal nodes appear continuously in the initial networking deployment, the center weight value is linearly increased; when two or more non-abnormal networking nodes appear continuously, the center weight value is linearly decreased.

9. The method of claim 8, wherein, The key link nodes and the heterogeneous data flow intensive nodes are screened out in all the networking nodes; when the key link nodes and the heterogeneous data flow intensive nodes are marked as abnormal nodes, the center weight value is adjusted preferentially.

Images