Method and System for Dynamically Adjusting Service Deployment in Road Networks Based on Agent Self-Organization

Abstract

The present invention provides a method and system for dynamically adjusting service deployment in a road network based on Agent self-organization. According to a preset decision sequence, for each service on each roadside unit (RSU) in the road network, based on its own information and neighbor information, its status is decided and adjusted. The decision adjustment basis includes: simulating the execution of requests in the opposite state of the current state of the target service, calculating the decision factor, and adjusting the status according to the magnitude of the decision factor value. For the extinction decision, additional judgments are made using the number of deployed services of the same type, the timeout rate and the normal processing rate of the service itself. The present invention focuses on the self-organization and dynamic adjustment of services, aiming to automatically adjust the cloning and extinction of services by using local information such as its own information and neighbor information, reducing manual operations and realizing the self-organization and dynamic optimization deployment of services.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent transportation, and specifically, to a method and system for dynamically adjusting service deployment in a road network based on Agent self-organization. Background Art

[0002] Vehicle-road collaborative V2X is an intelligent transportation system that uses roadside units and in-vehicle systems as a basis for information collection, transmits information through wireless communication technology, stores and makes decisions through cloud control technology, and finally realizes information interaction and sharing among people, vehicles, and roads. In the edge computing mode, vehicle-road collaboration needs to deploy various services on roadside units, such as services like speeding detection and non-pedestrian yielding detection in non-site law enforcement. Each service consumes a certain amount of resources (CPU, memory, etc.) and has specific functions, and there are also different time delay constraints in processing corresponding requests. Currently, in a static environment (a certain number of vehicles and a certain vehicle speed), various algorithms can be used to achieve optimal service deployment.

[0003] Patent document CN105847326A (application number: CN201610144035.1) discloses a vehicle-connected cloud resource dynamic deployment system, including a resource integration module, a resource management module, and a resource maintenance module. The resource integration module is used to complete the discovery, classification, and encapsulation of VC resources. The resource management module is used to uniformly schedule the resource pool to achieve information sharing and business collaboration. The resource integration module and the resource management module are constructed by a four-level node mechanism. The four-level node mechanism is, from top to bottom, a VC resource management center, a roadside unit, an agent node, and a user. The four-level node mechanism improves the scalability and robustness of the system, while reducing the complexity and dependence of the entire system, and avoiding the impact on other nodes caused by the downtime of a single node.

[0004] In real life, a static environment does not exist. The number of vehicles and the vehicle speed are variables that change dynamically. Therefore, the number of service requests will also change accordingly, and the service deployment obviously needs to be adjusted to a certain extent. Therefore, according to the request information of each service on the roadside unit, providing a service dynamic adjustment method based on Agent self-organization is a major technical problem that those skilled in the art urgently need to solve. Summary of the Invention

[0005] Aiming at the defects in the prior art, the purpose of the present invention is to provide a method and system for dynamically adjusting service deployment in a road network based on Agent self-organization.

[0006] According to the method for dynamically adjusting service deployment in a road network based on Agent self-organization provided by the present invention, for each service on each roadside unit (RSU) on the road network, according to the preset decision sequence, its state is decided and adjusted based on its own information and neighbor information. The decision adjustment basis includes: simulating the execution of requests in the opposite state of the current state of the target service, calculating the decision factor, and adjusting the state according to the value of the decision factor. For the extinction decision, additional judgments are made using the number of deployed services of the same type, the timeout rate, and the normal processing rate of the service itself.

[0007] Preferably, assume that the target service is in an undeployed state, and simulate the processing of all requests received within one cycle to obtain the number of timeout requests noResponse_new, the number of normally processed requests processed_new, the average request processing delay averageDelay_new, and the utilization rate useRatio_new of the target service after simulation;

[0008] Calculate the obtained data from the simulation and the data obtained from the actual execution of requests to obtain the influence value score_self on itself, use the simulation to obtain the influence value score_neighbor on neighbors, and finally add score_self and score_neighbor to obtain the decision factor alpha;

[0009] The calculation formula for score_self is:

[0010]

[0011] Wherein, processed, noReponse, averageDelay, and useRatio represent the number of normally processed requests, the number of timeout requests, the average request processing delay, and the service utilization rate obtained by the target service when processing requests in the real state, and total represents the total number of requests received by the target service within this cycle.

[0012] Preferably, if the target service was originally in an undeployed state, the decision factors of all its neighbors are obtained in the same way, and it is judged whether the decision factor value of the target service is greater than the decision factor values of all other neighbors. If so, the target service is cloned; if not, it remains in the undeployed state;

[0013] If the target service was originally in a deployed state, first judge whether the number of services of the same type on the current road network is greater than 1. If not, it remains in the deployed state; if so, continue to obtain the timeout rate noResponse_R and the normal processing rate processed_R of the target service within one cycle. The expressions are:

[0014]

[0015]

[0016] If noReponse_R ≥ 0.5 or processed_R ≤ 0.5, then maintain the deployed state; otherwise, use the decision factor alpha for judgment. If alpha < 1.0, then terminate the target service; otherwise, maintain the deployed state.

[0017] Preferably, the deployment of the service is evaluated after decision adjustment every T time, and the calculation method is as follows:

[0018]

[0019]

[0020]

[0021]

[0022] Among them, serviceCount is the total number of services; processedRatio is the average normal response rate of the service; noResponseRatio is the average response timeout rate of the service; averageDelay is the average normal processing delay of the service; averageUseRatio is the average service utilization rate; RSUS represents the set of all RSUs on the road network; service represents the set of all service types; len represents the size of the obtained object; process_queue and noResponse_queue respectively represent the normal processing queue and the timeout queue; total represents the total number of requests on a single RSU within T time; count represents the number of RSUs on the road network that have deployed service s; total_process_time represents the total delay spent on processing requests by service s in one cycle on a single RSU; total_process_count represents the total number of requests processed by service s in one cycle on a single RSU; use_time represents the delay actually used for computing and processing by a single RSU in one cycle; checkInterval represents the adjustment period T.

[0023] Preferably, the calculation formula for the decision factor alpha of the target service is as follows:

[0024]

[0025] Among them, score_self is the impact value on itself when the target service decision is deployed; N represents the set of RSUs affected among the RSU neighbors; r is the RSU affected among the RSU neighbors; score_neighbor is the impact value of the decision on the neighbors; when the decision factor value of the target service is the largest among its neighbors, its decision is adjusted to the opposite state, that is, cloning is performed, otherwise it remains in the undeployed state.

[0026] According to the system for dynamically adjusting service deployment in a road network based on Agent self-organization provided by the present invention, according to a preset decision sequence, for each service on each roadside unit RSU on the road network, based on its own information and neighbor information, its state is decision-adjusted. The decision adjustment basis includes: simulating the execution of requests in the opposite state of the current state of the target service, calculating the decision factor, and adjusting the state according to the magnitude of the decision factor value. For the extinction decision, additional judgments are made using the number of deployed services of the same type, the timeout rate and the normal processing rate of the service itself.

[0027] Preferably, assuming that the target service is in the undeployed state, all requests received within one cycle are simulated and processed once to obtain the number of timeout requests noResponse_new, the number of normally processed requests processed_new, the average request processing delay averageDelay_new, and the utilization ratio useRatio_new of the target service after simulation;

[0028] The data obtained by simulation is calculated with the data obtained from the actual execution of requests, and then the impact value score_self on itself is obtained. The impact value score_neighbor on the neighbors is obtained by simulation. Finally, the decision factor alpha is obtained by adding score_self and score_neighbor;

[0029] The calculation formula for score_self is:

[0030]

[0031] Among them, processed, noReponse, averageDelay, and useRatio represent the number of normally processed requests, the number of timeout requests, the average request processing delay, and the service utilization ratio obtained by processing requests in the real state of the target service, and total represents the total number of requests received by the target service within this cycle.

[0032] Preferably, if the target service is originally in an undeployed state, the decision factors of all its neighbors are obtained in the same way, and it is judged whether the value of the decision factor of the target service is greater than the decision factors of all other neighbors. If so, the target service is cloned; if not, it remains in the undeployed state.

[0033] If the target service is originally in a deployed state, first judge whether the number of services of the same type on the current road network is greater than 1. If not, it remains in the deployed state; if so, continue to obtain the timeout rate noResponse_R and the normal processing rate processed_R of the target service in one cycle. The expressions are as follows:

[0034]

[0035]

[0036] If noReponse_R≥0.5 or processed_R≤0.5, it remains in the deployed state; otherwise, it is judged using the decision factor alpha. If alpha<1.0, the target service is terminated; otherwise, it remains in the deployed state.

[0037] Preferably, every T time, the deployment of the service is evaluated after the decision adjustment. The calculation method is as follows:

[0038]

[0039]

[0040]

[0041]

[0042] Among them, serviceCount is the total number of services; processedRatio is the average normal service response rate; noResponseRatio is the average service response timeout rate; averageDelay is the average normal service processing delay; averageUseRatio is the average service utilization rate; RSUS represents the set of all RSUs on the road network; service represents the set of all service types; len represents the size of the obtained object; process_queue and noResponse_queue represent the normal processing queue and the timeout queue respectively; total represents the total number of requests on a single RSU within time T; count represents the number of RSUs on the road network that have deployed service s; total_process_time represents the total delay spent on processing requests by service s in one cycle on a single RSU; total_process_count represents the total number of requests processed by service s in one cycle on a single RSU; use_time represents the delay that a single RSU actually uses for computing and processing within one cycle; checkInterval represents the adjustment period T.

[0043] Preferably, the calculation formula of the decision factor alpha of the target service is as follows:

[0044]

[0045] Among them, score_self is the influence value on itself when the target service is decided to be deployed; N represents the set of RSUs affected among the RSU neighbors; r is the RSU affected among the RSU neighbors; score_neighbor is the influence value of the decision on the neighbors; when the decision factor value of the target service is the largest among its neighbors, its decision is adjusted to the opposite state, that is, cloning is performed, otherwise it remains in the undeployed state.

[0046] Compared with the prior art, the present invention has the following beneficial effects:

[0047] The present invention focuses on the self-organizing dynamic adjustment of services, aiming to automatically adjust the cloning and extinction of services by using local information such as its own information and neighbor information, reducing manual operations and realizing the self-organizing dynamic optimal deployment of services. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] By reading the detailed description of the non-restrictive embodiments with reference to the following drawings, other features, objectives and advantages of the present invention will become more obvious:

[0049] Figure 1 It is a schematic diagram of making a cloning decision on a service in the present invention;

[0050] Figure 2 Schematic diagram for service extinction decision-making in the present invention;

[0051] Figure 3 Schematic diagram for decision-making on all services in the road network in the present invention. Detailed implementation manners

[0052] The present invention will be described in detail below in conjunction with specific embodiments. The following embodiments will help those skilled in the art to further understand the present invention, but do not limit the present invention in any form. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several changes and improvements can still be made. These all fall within the protection scope of the present invention.

[0053] Embodiment:

[0054] The present invention provides a method for dynamically adjusting service deployment in a road network based on Agent self-organization, including: according to the pre-determined decision-making order, for each service on each RSU in the road network (such as overspeed detection service, non-yielding to pedestrians detection service, etc. in the field of non-site law enforcement), automatically make decision adjustments to its state according to its own information and neighbor information, so as to realize the optimized deployment of services adapting to the traffic flow density distribution on the road network.

[0055] The indicators for evaluating the quality of service deployment include:

[0056] Noun Meaning serviceCount Total number of services processedRatio Average normal service response rate noResponseRatio Average service response timeout rate averageDelay Average normal service processing delay averageUseRatio Average service utilization rate

[0057] Evaluate the service deployment after decision adjustment every T time. Their specific calculation methods are as follows (the number of services serviceCount can directly count whether all services on each RSU in the road network are deployed):

[0058]

[0059]

[0060]

[0061]

[0062] Among them, RSUS represents the set of all RSUs on the road network; service represents the set of all service types; len represents the size of the object to be obtained; process_queue and noResponse_queue represent the normal processing queue and the timeout queue respectively; total represents the total number of requests on a single RSU within time T; count represents the number of RSUs on the road network that have deployed service s; total_process_time represents the total delay spent on processing requests in one cycle of service s on a single RSU; total_process_count represents the total number of requests processed by service s on a single RSU in one cycle; use_time represents the delay that a single RSU actually uses for computing and processing within one cycle; checkInterval represents the adjustment period T.

[0063] The experimental scenario of this invention is as follows: A vehicle sends a service request of a certain type to the neighboring RSU_A. After the request is received by RSU_A, it is judged whether the service itself is deployed. If it has been deployed, the request is put into the request queue (RQ) of this service on RSU_A and waits to be processed; if it has not been deployed, the request is forwarded to RSU_B that has deployed this service, and the result is returned to RSU_A after processing. The processing results of the request include two types: one is timeout and the request cannot be normally responded to, and the requests with this result are put into the timeout queue (NQ) of this service on RAU_A; the other is normal response, and the processing result is returned to the vehicle, and the requests with this result are put into the normal processing queue (PQ) of this service on RSU_A. Every T time, all services in the road network are decision-adjusted to adapt to the traffic flow density distribution and achieve optimized deployment.

[0064] When making decision adjustments, in accordance with the pre-determined decision order (taking the road network as a graph, RSUs as nodes, starting from any node, and performing depth-first traversal to obtain the decision order), decisions are made for each service on each RSU.

[0065] When making decisions on service A on RSU_A, there are two situations: service A is deployed on RSU_A and service A is not deployed on RSU_A. The decision-making methods for the two situations are different. If service A is not deployed, the decision result is to clone or keep the undeployed state unchanged; otherwise, the decision result is to perish or keep the deployed state unchanged.

[0066] If service A is in the undeployed state, the method of decision adjustment is as Figure 1As shown below. First, obtain the decision factor value alpha of Service A. Assuming that Service A is in its opposite state (deployed), it will have an impact on Service A on itself and some of the neighbors of RSU_A. For itself, the requests originally received by Service A on RSU_A would be forwarded to the nearby RSUs with this service for processing, resulting in a forwarding delay. If Service A on RSU_A is in the deployed state, there will be no forwarding delay, which will reduce the timeout rate. Therefore, using the request information received by Service A on RSU_A within time T, re-simulate the request processing for a T cycle. The key is that there is no forwarding delay when processing requests, and then obtain the number of timeout requests (noResponse_new), the number of requests processed normally (processed_new), the average request processing delay (averageDelay_new), and the utilization rate of Service A (useRatio_new) after simulation. Calculate the impact factor value score_self of the decision on itself using the data obtained from real execution and the data obtained from simulation. The calculation formula is as follows:

[0067]

[0068] Among them, processed, noReponse, averageDelay, and useRatio respectively represent the number of requests processed normally, the number of timeout requests, the average request processing delay, and the service utilization rate obtained when Service A processes requests in the real state. total represents the total number of requests received by Service A during this cycle.

[0069] After calculating the impact value score_self of the decision that Service A on RSU_A is deployed on itself, consider the impact of this decision on Service A on some of the neighbors of RSU_A. These neighbors have one thing in common, that is, they do not deploy Service A themselves, and none of their neighbors deploy Service A either. Therefore, if RSU_A deploys Service A, for these neighbors, the forwarding distance will be reduced to the distance from them to RSU_A when receiving requests of Service A type, the forwarding delay becomes smaller, and evaluation indicators such as the timeout rate will be improved accordingly. Therefore, simulate these neighbors and re-simulate the execution of all the requests they received during this cycle. The key is that for these requests, the forwarding delay of requests of Service A type will be reduced. Calculate the impact value score_neighbor of the decision using the obtained simulation data and real execution data.

[0070] The calculation formula for the decision factor alpha of Service A on RSU_A is as follows:

[0071]

[0072] Among them, N represents the set of RSU_A neighbors that are affected. In the same way, the decision factor values of all neighbors of RSU_A for service A are calculated. When the decision factor value of service A on RSU_A is the largest among its neighbors, its decision is adjusted to the opposite state (deployed), that is, cloning is performed, otherwise it remains in the undeployed state.

[0073] If service A is in the deployed state, the method of decision adjustment is as Figure 2 shown. First, it is judged whether the number of deployed service A on the entire road network is greater than 1. If it is not greater than 1, it means that the service A on RSU_A is the last service of type A at present and is not allowed to die. Secondly, since making decisions based on the relationship between the decision factor and the threshold will cause certain misjudgments (for example: if service A is overloaded, even if it is killed, the effect will not become too bad, resulting in too small a decision factor and then wrongly killing it), therefore, additional load judgment is needed. The timeout rate and normal processing rate of service A are obtained through the number of timeout requests (noResponse), the number of normally processed requests (processed), and the total number of requests (total) obtained by service A processing requests within a cycle. The formulas are as follows:

[0074]

[0075]

[0076] When noResponse_R >= 0.5 or processed_R <= 0.5, it means that service A is too busy in this cycle and is not allowed to die. Finally, the decision factor alpha of service A and the threshold are used for decision adjustment. As Figure 3 , assuming that service A is in its opposite state (undeployed), it will have an impact on itself and some neighbors. For itself, the original requests of service A type will be forwarded to the RSU with this type of service for processing, and there will be a forwarding delay. The impact value score_self on itself is obtained through simulation execution; for neighbors, there are some neighbors. These neighbors have in common that they do not deploy service A themselves, and among their neighbors, except for RSU_A that deploys service A, no other RSU deploys service A. For these neighbors, if the service A on RSU_A dies, the A service requests they receive will be forwarded to a farther RSU, and the forwarding delay will increase. The impact value score_neighbor on these neighbors is obtained through simulation execution. The decision factor alpha of service A on RSU_A is obtained through score_self and score_neighbor. When alpha < 1.0, the decision of this service is adjusted to the opposite state (undeployed), that is, killed, otherwise, the existing deployed state is maintained.

[0077] Those skilled in the art know that in addition to implementing the systems, devices, and their respective modules provided by the present invention in the form of pure computer-readable program code, it is entirely possible to logically program the method steps so that the systems, devices, and their respective modules provided by the present invention are implemented in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, embedded microcontrollers, etc. to achieve the same program. Therefore, the systems, devices, and their respective modules provided by the present invention can be considered as a kind of hardware component, and the modules included therein for implementing various programs can also be regarded as the structures within the hardware component; the modules for implementing various functions can also be regarded as either software programs for implementing the method or the structures within the hardware component.

[0078] The specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the above specific embodiments, and those skilled in the art can make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. Without conflict, the embodiments of the present application and the features in the embodiments can be combined with each other arbitrarily.

Claims

1. A method for dynamically adjusting service deployment in a road network based on Agent self-organization, characterized in that According to the preset decision sequence, for each service on each roadside unit (RSU) on the road network, based on its own information and neighbor information, the state is adjusted by making decisions. The decision adjustment basis includes: simulating the execution of requests in the opposite state of the current state of the target service, calculating the decision factor, and adjusting the state according to the magnitude of the decision factor value. For the termination decision, additional judgments are made using the number of deployed services of the same type, the timeout rate, and the normal processing rate of the service itself. Assume that the target service is in the undeployed state. Simulate the processing of all requests received within one cycle of it once to obtain the number of timeout requests noResponse_new, the number of normally processed requests processed_new, the average request processing delay averageDelay_new, and the utilization ratio useRatio_new of the target service after simulation. Calculate the data obtained from the simulation and the data obtained from the actual execution of requests to obtain the influence value score_self on itself, use the simulation to obtain the influence value score_neighbor on neighbors, and finally obtain the decision factor alpha by adding score_self and score_neighbor. The calculation formula for score_self is: Among them, processed, noReponse, averageDelay, and useRatio represent the number of normally processed requests, the number of timeout requests, the average request processing delay, and the service utilization ratio obtained by the target service when processing requests in the actual state, and total represents the total number of requests received by the target service within this cycle. If the target service was originally in the undeployed state, use the same method to obtain the decision factors of all its neighbors, and judge whether the decision factor value of the target service is greater than the decision factor values of all other neighbors. If so, clone the target service; if not, keep it in the undeployed state. If the target service was originally in the deployed state, first judge whether the number of services of the same type on the current road network is greater than 1. If not, keep it in the deployed state; if so, continue to obtain the timeout rate noResponse_R and the normal processing rate processed_R of the target service within one cycle. The expressions are: If noReponse_R ≥ 0.5 or processed_R ≤ 0.5, keep it in the deployed state; otherwise, use the decision factor alpha for judgment. If alpha < 1.0, terminate the target service; otherwise, keep it in the deployed state.

2. The method for dynamically adjusting service deployment in a road network based on Agent self-organization according to claim 1, characterized in that, Evaluate the deployment of the service at intervals of T time after the decision adjustment. The calculation method is: Among them, serviceCount is the total number of services; processedRatio is the average normal service response rate; noResponseRatio is the average service response timeout rate; averageDelay is the average normal service processing delay; averageUseRatio is the average service utilization rate; RSUS represents the set of all RSUs on the road network; service represents the set of all service types; len represents the size of the obtained object; process_queue and noResponse_queue represent the normal processing queue and the timeout queue respectively; total represents the total number of requests on a single RSU within time T; count represents the number of RSUs on the road network that have deployed service s; total_process_time represents the total delay spent on processing requests by service s in one cycle on a single RSU; total_process_count represents the total number of requests processed by service s in one cycle on a single RSU; use_time represents the delay that a single RSU actually spends on computing and processing within one cycle; checkInterval represents the adjustment period T.

3. A system for dynamically adjusting service deployment in a road network based on Agent self-organization, characterized in that, According to the preset decision sequence, for each service on each roadside unit RSU on the road network, based on its own information and neighbor information, make a decision adjustment on its state. The decision adjustment basis includes: simulating the execution of requests in the opposite state of the current state of the target service, calculating the decision factor, and adjusting the state according to the value of the decision factor. For the extinction decision, make an additional judgment using the number of deployed services of the same type, the timeout rate and the normal processing rate of the service itself. Assume that the target service is in the undeployed state, simulate the processing of all requests received within one cycle once, and obtain the number of timeout requests noResponse_new, the number of normally processed requests processed_new, the average request processing delay averageDelay_new and the utilization rate useRatio_new of the target service after simulation. Calculate the data obtained from the simulation and the data obtained from the actual execution of requests, and then obtain the influence value score_self on itself. Use the simulation to obtain the influence value score_neighbor on the neighbor. Finally, add score_self and score_neighbor to obtain the decision factor alpha. The calculation formula for score_self is: Among them, processed, noReponse, averageDelay, and useRatio represent the number of normally processed requests, the number of timeout requests, the average request processing delay, and the service utilization rate obtained by the target service when processing requests in the real state. total represents the total number of requests received by the target service within this cycle. If the target service is originally in an undeployed state, the decision factors of all its neighbors are obtained in the same way, and it is judged whether the value of the decision factor of the target service is greater than the decision factors of all other neighbors. If so, the target service is cloned; if not, it remains in the undeployed state. If the target service is originally in a deployed state, first judge whether the number of services of the same type on the current road network is greater than 1. If not, it remains in the deployed state. If so, continue to obtain the timeout rate noResponse_R and the normal processing rate processed_R of the target service within one cycle. The expressions are as follows: If noReponse_R≥0.5 or processed_R≤0.5, it remains in the deployed state. Otherwise, it is judged using the decision factor alpha. If alpha<1.0, the target service is terminated; otherwise, it remains in the deployed state.

4. The system for dynamically adjusting service deployment in a road network based on Agent self-organization according to claim 3, characterized in that, The deployment of the service is evaluated every T time after the decision adjustment. The calculation method is as follows: Among them, serviceCount is the total number of services; processedRatio is the average normal response rate of the service; noResponseRatio is the average service response timeout rate; averageDelay is the average normal processing delay of the service; averageUseRatio is the average service utilization rate; RSUS represents the set of all RSUs on the road network; service represents the set of all service types; len represents the size of the obtained object; process_queue and noResponse_queue represent the normal processing queue and the timeout queue respectively; total represents the total number of requests on a single RSU within T time; count represents the number of RSUs on the road network that have deployed service s; total_process_time represents the total delay spent on processing requests by service s on a single RSU within one cycle; total_process_count represents the total number of requests processed by service s on a single RSU within one cycle; use_time represents the delay actually used for calculation and processing by a single RSU within one cycle; checkInterval represents the adjustment period T.

Images