Search

CN-122027569-A - Space-time behavior sensing and dynamic shaping method for SOME/IP service flow

CN122027569ACN 122027569 ACN122027569 ACN 122027569ACN-122027569-A

Abstract

The invention discloses a space-time behavior sensing and dynamic shaping method for SOME/IP service flows, which comprises the steps of collecting SOME/IP reports Wen Jingxiang by-pass of a main control platform of a vehicle-mounted network controller, analyzing and extracting space-time multidimensional features, constructing a service granularity flow sample set based on a sliding window, completing nuclear density estimation through self-adaptive grid discretization and frequency domain acceleration convolution, deeply characterizing service flow probability density distribution, quantifying space-time communication behaviors of the service flows by comprehensive information entropy, carrying out on-line comparison by means of self-adaptive dual-entropy threshold decision, precisely triggering a hierarchical shaping strategy, driving an ECU at a release end to execute in real time by means of a standard interface, fusing a bottom physical state to construct closed loop recursive feedback for dynamic correction, and setting a severe boundary constraint and deterministic failure rollback mechanism based on state deviation. The invention realizes SOME/IP service level accurate management and control, low time delay shaping and vehicle-standard level safety guarantee, does not need to modify a bottom protocol stack, and builds a data-driven dynamic flow control paradigm for a vehicle-mounted network.

Inventors

  • WANG YUEFEI
  • ZHAO RUIHUA
  • Yi Zhejin
  • Zhou Yaman
  • YUAN YICHEN
  • SUN YONGQING

Assignees

  • 合肥工业大学

Dates

Publication Date
20260512
Application Date
20260414

Claims (9)

  1. 1. A space-time behavior perception and dynamic shaping method facing SOME/IP service flow is characterized by comprising the following steps: Step 1, a message collection module is deployed on a main control processing platform in a vehicle-mounted network domain controller, a central gateway or an area controller, and a bypass mode is adopted by the message collection module to passively receive a SOME/IP report Wen Jingxiang copy; Step 2, the message collection module performs verification analysis on the received SOME/IP message mirror image copy, extracts space-time multidimensional feature vectors used for representing semantics and communication behaviors, and aggregates the space-time multidimensional feature vectors based on a sliding time window mechanism to construct a service granularity flow sample set; Step 3, performing kernel density estimation on the flow sample set, namely performing self-adaptive gridding discrete processing on a feature space formed by the space-time multidimensional feature vectors, converting the kernel density estimation into weighted convolution calculation for service flows, and constructing non-parameter probability density distribution for representing SOME/IP service flows; Step 4, defining SOME/IP service flow information entropy based on the non-parameter probability density distribution, performing hierarchical sampling-based numerical approximate calculation on the SOME/IP service flow information entropy, and obtaining service flow comprehensive information entropy in the current sliding time window To quantify the complexity and uncertainty of the spatio-temporal communication behavior of the copy of the SOME/IP report Wen Jingxiang; Step 5, based on the self-adaptive dual-entropy threshold decision mechanism, the service flow comprehensive information entropy is calculated Comparing with the self-adaptive double-entropy threshold on line to judge the running state of the SOME/IP service flow in the sliding time window, and triggering the corresponding hierarchical dynamic shaping strategy according to the predefined rule; Step 6, the hierarchical dynamic shaping strategy is issued to the issuing end ECU through a standard communication interface or an event notification mechanism, and the issuing end ECU executes the hierarchical dynamic shaping strategy in a subsequent communication period; Step 7, continuously monitoring the SOME/IP service flow after executing the dynamic shaping strategy, constructing a closed loop recursive feedback mechanism by combining the running state deviation, and dynamically adjusting core control parameters for dynamic sensing, decision making and dynamic shaping strategy execution on line; and 8, based on a preset safety boundary constraint and a deterministic failure rollback mechanism, when the entropy safety judgment index is detected to exceed a preset safety boundary threshold, automatically stopping dynamic shaping adjustment of the SOME/IP service flow, enabling the SOME/IP service flow to be restored to a predefined default configuration state, continuously monitoring the time-space communication behavior during rollback, and restarting the dynamic shaping flow when the state is stabilized within the safety boundary.
  2. 2. The method for space-time behavior perception and dynamic shaping for SOME/IP service flow according to claim 1, wherein the specific process of extracting space-time multidimensional feature vectors and constructing a flow sample set in step 2 comprises the following steps: Step 2.1, performing header integrity check and protocol field analysis on the captured copy of the SOME/IP report Wen Jingxiang, and extracting space dimension core identification features for identifying services and communication semantics, wherein the space dimension core identification features comprise a service identifier, an instance identifier, an event group identifier and a message type; Step 2.2, extracting the load length of the SOME/IP report Wen Jingxiang copy, calculating the arrival time interval of the adjacent messages and obtaining the time stamp by combining the message capturing time, combining the space dimension features and the time dimension features, and constructing a multidimensional feature vector for jointly representing the space-time communication behavior of the SOME/IP service flow ; Wherein, the In order to provide a service identifier for the service, In order to provide an instance identifier, In the event of an event identifier being identified, In order to provide an event group identifier, As a result of the type of message, For the length of the load, For the inter-arrival of adjacent messages, Is a time stamp; step 2.3, adopting a sample maintenance strategy based on increment updating, inserting the newly constructed multidimensional feature vector into the set, and removing the expired feature vector beyond the sliding time window to update online and construct a flow sample set : Wherein, the And t is the discrete time index of the current sliding time window.
  3. 3. The method for space-time behavior awareness and dynamic shaping for SOME/IP service flows according to claim 2, wherein the non-parametric probability density distribution expression in the step 3 is as follows: as a function of the kernel, For the adaptive core bandwidth associated with the samples, Time decay weight coefficients set for each sample within the current sliding time window, As a dimension of the feature space, The feature mapping function is used for carrying out joint mapping on the service identifier, the message type and the time domain feature in the multidimensional feature vector; The specific steps for performing the nuclear density estimation include: Step 3.1, determining the self-adaptive core bandwidth, namely determining the self-adaptive core bandwidth corresponding to different SOME/IP service sub-streams Wherein the initial value of the adaptive kernel bandwidth Determined by the following formula: in order to serve the corresponding feature space dimension, For the number of samples of the service within the sliding time window, For the statistical scale parameter of the corresponding service sample, Is a preset adjustment factor; In the online running process of the system, the self-adaptive core bandwidth is smoothly updated according to the following steps: in order to adaptively update the coefficients, A target bandwidth estimated according to the current service flow space-time behavior; Step 3.2, based on the preset objective statistical characteristic threshold, performing adaptive switching between different types of kernel functions, namely when the feature space dimension of the service sub-stream is monitored When the variance of the arrival time interval of the adjacent messages is smaller than a preset fluctuation threshold, the kernel function adopts the following formula: Is that A normalization constant in dimensional space; Is an indication function; When feature space dimension of service substream is monitored When the variance is larger than or equal to the preset fluctuation threshold, the method is switched to adopting the following formula: A bandwidth scaling factor associated with the service type and the message type; step 3.3 includes a weighted convolution calculation that converts the kernel density estimate into a service-oriented stream, specifically including: step 3.3.1, performing self-adaptive grid discretization processing on the feature space according to the bandwidth parameters corresponding to each service, and mapping the flow sample set into discrete count signals [m]: Is a characteristic grid cell associated with service s, discrete count signal Composing a system discretized representation Is a service level component of (1); step 3.3.2, kernel function pairs in the same discrete space Sampling to obtain discrete kernel function Performing an accelerated convolution calculation to obtain a discretized probability density distribution : Wherein, the Represents a frequency domain conversion operator that maps the spatial domain discrete signal to the frequency domain to equate the convolution to a scalar multiplication, Representing its corresponding spatial domain inverse transform operator.
  4. 4. A method for spatio-temporal behavior awareness and dynamic shaping of a stream of SOME/IP services according to claim 3, characterized in that step 4 comprises: Step 4.1, based on the service communication characteristic of SOME/IP protocol, according to the joint identification of service identifier, instance identifier and event group identifier, making service granularity division on the collected message and making service granularity division on the flow sample set Sub-stream sample set of individual service instances: Representing the SOME/IP service sub-stream index, As a function of the feature map, Is the first The number of message samples in the sliding time window by the individual service substreams; step 4.2, defining service substream information entropy as a continuous integral form based on the service feature probability density function obtained by kernel density estimation for each service substream: (x) Is the first A conditional probability density function of the individual service substreams within a current sliding time window; step 4.3, dividing the multidimensional feature space into Individual feature subintervals Carrying out differential sampling point distribution according to the contribution degree of each subinterval to the information entropy, wherein the first is that Individual feature subintervals Distributed sample points The method meets the following conditions: step 4.4, at each characteristic subinterval In, in order to propose distribution Sampling to obtain a sample Weighting and correcting samples according to the relation between proposal distribution and corresponding service flow probability density function, and calculating importance weight of the samples : , For the preset proposal distribution, the method comprises the steps of, Is an adjustment factor associated with the service identifier and the message type; step 4.5, based on the extracted sampling points and the corresponding sample importance weights, performing numerical approximation calculation on the service substream information entropy: is the first The probability weights of the sub-intervals, Representing sample points Mapping to the probability density estimated value corresponding to the discrete grid; step 4.6, weighting and fusing the information entropy of each service sub-stream to obtain the comprehensive information entropy of the service stream in the main weight item : Wherein, the For the total number of service substreams currently being processed in parallel by the system, And (5) the weight coefficient of each service sub-stream is preconfigured.
  5. 5. The method for space-time behavior awareness and dynamic shaping for SOME/IP service flows according to claim 4, wherein said step 5 comprises: step 5.1, in the system initialization stage, performing distributed clustering on entropy indexes based on a historical data set, and determining an initial low entropy threshold value With an initial high entropy threshold ; Based on the recent entropy sequence { within the sliding window ,..., And adopting a progressive drift updating mechanism for the low entropy threshold and the high entropy threshold respectively: for the threshold value to be updated with the rate coefficient, A time window length that is a recent sequence; And (3) with Respectively representing the fractional operators of the minimum fractional parameter and the maximum fractional parameter; step 5.2, the current entropy value And updated low entropy threshold And a high entropy threshold Performing on-line comparison; If it is Triggering a load-shedding shaping strategy; If it is Triggering an enhanced shaping strategy; otherwise, maintaining a standard SOME/IP transport mechanism; And 5.3, mapping and executing a load-shedding shaping strategy, namely carrying out load-shedding shaping on the service from three behavior dimensions of service transmission scheduling weight, event triggering rhythm and message organization mode, wherein the method specifically comprises the following steps: Step 5.3.1, reducing the scheduling weight of the corresponding service in the communication scheduling in the service transmission scheduling weight dimension The update method is based on the following formula: in order to be able to change the rate of entropy, For the reference scheduling weights corresponding to the SOME/IP services, Weight adjusts the mapping function when Or (b) When the preset interval is exceeded, outputting a weight adjusting factor smaller than 1; Step 5.3.2, in the event triggered rhythms dimension, adjust its event triggered rhythms for periodic SOME/IP event notifications And a batch merging and sending mechanism is started in a communication scheduling layer, and the adjustment mode is as follows: ) Wherein, the A rhythm is triggered for a reference event of a corresponding service, For a monotonic mapping function related to service event transmission characteristics, Adjusting smoothing coefficients for the rhythms; step 5.3.3, in the dimension of the message organization mode, adopting load increment compression transmission based on front-back message difference to the continuous or adjacent event notification messages; Step 5.3.4, for the key service related to the functional safety, keeping the preset time delay constraint and the minimum scheduling weight unchanged; and 5.4, mapping and executing an enhanced shaping strategy, namely enhancing and protecting the related service flow from two behavior dimensions of service transmission scheduling weight and communication isolation.
  6. 6. The method for space-time behavior awareness and dynamic shaping for a SOME/IP service flow according to claim 5, wherein step 5.4 specifically comprises: Step 5.4.1, in the dimension of the service transmission scheduling weight, the scheduling priority or the scheduling weight of the corresponding service in the communication scheduling is improved The update mode satisfies the following formula: In order to enhance the adjustment coefficient, The updating mode is used for enabling the scheduling weight to be not smaller than the reference scheduling weight in the enhanced control scene; And 5.4.2, synchronously starting a bandwidth protection or logic isolation mechanism, limiting speed, isolating or restraining priority for abnormal or burst service flows, and triggering a corresponding abnormal state indication for reporting or recording the abnormal operation state of the service flows.
  7. 7. The method for space-time behavior awareness and dynamic shaping for a stream of SOME/IP services according to claim 6, wherein said step 6 comprises: And issuing a hierarchical dynamic shaping strategy to an application layer of the issuing-end ECU through an AUTOSAR ara: com standard communication interface or vsomeip event notification mechanism so as to drive an application program of the issuing-end ECU to finish dynamic adjustment of the sending behavior of the corresponding SOME/IP service on the premise of not modifying underlying communication protocol stack logic.
  8. 8. The method for space-time behavior awareness and dynamic shaping for a SOME/IP service flow according to claim 7, wherein the closed-loop feedback in step 7 includes: step 7.1, continuously collecting SOME/IP service flow after executing dynamic shaping strategy in the vehicle-mounted Ethernet communication path to obtain bandwidth occupation End-to-end delay Link load And constructs a service running state vector: to perform subsequent dynamic adjustment decisions based on the service operational state vector; step 7.2, entropy-driven self-adaptive closed-loop adjustment, namely entropy based on service flow comprehensive information Entropy with target reference Defining an entropy adjustment error term: based on the entropy adjustment error term, the entropy change rate and the deviation of the service running state vector, the composite adjustment parameter vector used for representing the service flow perception and state is subjected to recursive adjustment: Wherein, the For a gain mapping vector matching the complex tuning parameter vector dimension, For a gain matrix with state error mapped to the tuning parameter space, Is a preset reference state vector; The composite adjustment parameter vector For the collective characterization of the core control parameters which can be adjusted in an online self-adaptive manner in the preamble step, The internal dimension component mapping relationship is defined by the following equation: Wherein, the space-time behavior perception parameters Decision making class parameters Shaping execution class parameters The system acquires updated by recursion And decoupling each dimension component and respectively substituting the dimension components into the corresponding calculation step of the next time window, thereby constructing a data-driven global closed loop.
  9. 9. The method for space-time behavior awareness and dynamic shaping for a SOME/IP service flow according to claim 8, wherein the boundary constraint and failure rollback mechanism in step 8 further comprises: step 8.1, constructing a safety boundary function, namely aiming at the service flow comprehensive information entropy calculated in the sliding time window Constructing entropy security decision index for quantifying current communication state deviation degree : Wherein, the And (3) with The standard deviation and the average value of the service flow comprehensive information entropy obtained by statistics of the corresponding SOME/IP service flow in the historical stable operation stage are respectively, Indicating the rate of change of the entropy, For a preset rate term weight coefficient, Sampling time intervals for entropy values; Step 8.2, entropy safety judgment index With a pre-configured safety boundary threshold And (3) performing real-time comparison, if: Judging that the running state of the SOME/IP service flow exceeds a preset safety range, immediately triggering a failure rollback mechanism, terminating a dynamic shaping adjustment process of the SOME/IP service flow, and forcing the SOME/IP service flow to recover to a predefined default transmission strategy configuration state; Step 8.3, in the failure rollback state of default transmission policy configuration, synthesizing information entropy for the service flow Continuous background monitoring and index calculation are carried out, if: and judging that the communication state is stable, releasing the invalid rollback state, and allowing the SOME/IP service flow to reenter the dynamic shaping flow.

Description

Space-time behavior sensing and dynamic shaping method for SOME/IP service flow Technical Field The invention relates to the technical field of vehicle-mounted communication and network optimization, in particular to a space-time behavior perception and dynamic shaping method for SOME/IP service flows. Background Intelligent networking automobiles and software-defined vehicle architectures are rapidly evolving, and in-vehicle electronic and electrical architectures are evolving gradually towards centralized, regional and central computing architectures based on automobile ethernet. The SOME/IP protocol is widely used for service discovery, event notification and remote procedure call in a vehicular Ethernet environment as an important component of the AUTOSAR Adaptive platform. But it also introduces a new set of challenges in practical engineering applications. Firstly, SOME/IP adopts a dynamic service discovery mechanism to issue service information in a periodical broadcast or multicast mode, the mechanism possibly causes problems of redundancy, discovery failure or network load surge and the like of service discovery messages under the conditions of inconsistent ECU starting time sequence, enlarged network scale or more service quantity, and simultaneously SOME/IP messages CAN have time delay jitter or even packet loss phenomena under high-load or multi-service concurrency scenes, and compared with CAN frames, SOME/IP messages carry more protocol header information, the problems of bandwidth occupation increase, network congestion risk improvement and the like are easily caused under high-frequency event notification or large-scale data distribution scenes, and meanwhile, an encryption, authentication or abnormal traffic suppression mechanism is not built in SOME/IP protocols, and under abnormal traffic or potential attack scenes, key service communication CAN be influenced if effective monitoring and regulation means are lacked. In summary, how to perform real-time sensing, adaptive analysis and dynamic shaping on the space-time communication behavior of the SOME/IP-based service in the vehicle-mounted Ethernet to give consideration to the communication efficiency, the real-time performance and the functional safety is still a technical problem to be solved in the current intelligent network-connected automobile communication system. Disclosure of Invention The space-time behavior perception and dynamic shaping method for SOME/IP service flows provided by the invention can at least solve one of the technical problems in the background technology. In order to achieve the above purpose, the present invention adopts the following technical scheme: A space-time behavior perception and dynamic shaping method facing SOME/IP service flow includes the following steps: Step 1, a message collection module is deployed on a main control processing platform in a vehicle-mounted network domain controller, a central gateway or an area controller, and a bypass mode is adopted by the message collection module to passively receive a SOME/IP report Wen Jingxiang copy; Step 2, the message collection module performs verification analysis on the received SOME/IP message mirror image copy, extracts space-time multidimensional feature vectors used for representing semantics and communication behaviors, and aggregates the space-time multidimensional feature vectors based on a sliding time window mechanism to construct a service granularity flow sample set; Step 3, performing kernel density estimation on the flow sample set, namely performing self-adaptive gridding discrete processing on a feature space formed by the space-time multidimensional feature vectors, converting the kernel density estimation into weighted convolution calculation for service flows, and constructing non-parameter probability density distribution for representing SOME/IP service flows; Step 4, defining SOME/IP service flow information entropy based on the non-parameter probability density distribution, performing hierarchical sampling-based numerical approximate calculation on the SOME/IP service flow information entropy, and obtaining service flow comprehensive information entropy in the current sliding time window To quantify the complexity and uncertainty of the spatio-temporal communication behavior of the copy of the SOME/IP report Wen Jingxiang; Step 5, based on the self-adaptive dual-entropy threshold decision mechanism, the service flow comprehensive information entropy is calculated Comparing with the self-adaptive double-entropy threshold on line to judge the running state of the SOME/IP service flow in the sliding time window, and triggering the corresponding hierarchical dynamic shaping strategy according to the predefined rule; Step 6, the hierarchical dynamic shaping strategy is issued to the issuing end ECU through a standard communication interface or an event notification mechanism, and the issuing end ECU executes the hierarchical dynam