Multi-server-side matching method and device for vehicle-mounted server

Abstract

The invention relates to the technical field of server matching, in particular to a multi-server-side matching method and device for a vehicle-mounted server. The method comprises the steps of determining a current bandwidth of a vehicle-mounted server according to a bandwidth attenuation coefficient corresponding to a position where a target vehicle is located, a reference bandwidth of the vehicle-mounted server, a current speed of the target vehicle and a preset bandwidth model of the vehicle-mounted server in response to a situation that the target vehicle is currently in a potential safety hazard state; the transmission bandwidth between the vehicle-mounted server and each external server side is obtained by combining each preset transmission model of the preset transmission model set; obtaining the prediction result return time of each external server side according to the transmission bandwidth between the vehicle-mounted server and each external server side; and determining the external server with the minimum prediction result return time as the external server currently matched with the vehicle-mounted server, and sending the processed data to the vehicle-mounted server for processing. According to the invention, the external server with the highest processing efficiency can be determined, and the driving safety is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of server matching, and in particular to a multi-server-end matching method and device for a vehicle-mounted server. Background Art

[0002] In recent years, with the rapid development of intelligent driving technology, vehicles have gradually evolved from single means of transportation to highly integrated intelligent terminals. The number of on-board sensors and data dimensions have grown exponentially, covering cameras, lidar, millimeter-wave radar, inertial navigation systems, etc., providing rich information for vehicle environmental perception and decision-making. However, the computing power of on-board servers is limited, and it is difficult to support the efficient processing of large amounts of data, nor is it difficult to support high-complexity algorithms, resulting in difficulty in meeting the fast and accurate response requirements of intelligent driving. In order to solve this problem, some solutions in the existing technology attempt to upload data to an external server (such as the cloud) for processing. However, for scenarios with multiple external server ends, matching different external server ends often results in different processing efficiencies. How to determine the external server end with the highest processing efficiency is an urgent problem to be solved. Summary of the Invention

[0003] The present invention aims to provide a multi-server matching method and device for an on-board server, so as to determine the external server with the highest processing efficiency and improve driving safety.

[0004] According to a first aspect of the present invention, a multi-server matching method for an in-vehicle server is provided, comprising the following steps: S100, in response to the target vehicle currently being in a safety hazard state, the current bandwidth of the onboard server is determined based on the bandwidth attenuation coefficient corresponding to the location of the target vehicle, the baseline bandwidth of the onboard server, the current speed of the target vehicle and the preset bandwidth model of the onboard server; the preset bandwidth model of the onboard server includes the relationship between the bandwidth attenuation coefficient, the baseline bandwidth, the current speed and the current bandwidth.

[0005] S200, obtaining the transmission bandwidth between the on-board server and each external server end according to the current bandwidth of the on-board server and each preset transmission model in the preset transmission model set; the preset transmission model set includes n preset transmission models, n is the number of external server ends, the i-th preset transmission model corresponds to the i-th external server end, and the i-th preset transmission model includes the relationship between the transmission bandwidth between the on-board server and the i-th external server end and the current bandwidth of the on-board server; i=1,2,…,n.

[0006] S300, obtain the predicted result return time of each external server end based on the transmission bandwidth between the vehicle-mounted server and each external server end, the predicted processing time corresponding to each external server end, the queuing time corresponding to each external server end, the amount of data to be processed by the vehicle-mounted server and the amount of data of the processing results.

[0007] S400: Determine the external server with the shortest prediction result return time as the external server currently matching the vehicle-mounted server, and send the data to be processed by the vehicle-mounted server to the external server with the shortest prediction result return time for processing.

[0008] According to a second aspect of the present invention, a multi-server-side matching device for an in-vehicle server is provided, the device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned multi-server-side matching method for the in-vehicle server when executing the computer program.

[0009] Compared with the prior art, the present invention has at least the following beneficial effects: When the target vehicle is in a state of potential safety hazard, the present invention obtains the predicted result return time of each external server end, determines the external server end with the shortest predicted result return time as the external server end currently matched with the vehicle-mounted server, and sends the data to be processed by the vehicle-mounted server to the external server end with the shortest predicted result return time for processing. This facilitates the vehicle-mounted server to quickly receive the processing result and then quickly execute the next action, thereby improving the driving safety of the target vehicle. The current bandwidth of the vehicle-mounted server is also determined based on the bandwidth attenuation coefficient corresponding to the location of the target vehicle and the current speed of the target vehicle. Furthermore, the transmission bandwidth between the vehicle-mounted server and each external server end is obtained based on different preset transmission models corresponding to each external server end, making the predicted result return time obtained for each external server end more accurate, thereby ensuring that the vehicle-mounted server can send the data to be processed to the external server that can receive the processing result the fastest, thereby ensuring the driving safety of the target vehicle. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0011] Figure 1 A flowchart of a multi-server matching method for an in-vehicle server provided in the first embodiment of the present invention; Figure 2This is a schematic diagram of the vehicle-mounted server and the external server provided in the first embodiment of the present invention. DETAILED DESCRIPTION

[0012] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts shall fall within the scope of protection of the present invention. Example

[0013] According to this embodiment, Figure 1 As shown, a multi-server matching method for an on-board server is provided, comprising the following steps: S100, in response to the target vehicle currently being in a safety hazard state, the current bandwidth of the onboard server is determined based on the bandwidth attenuation coefficient corresponding to the location of the target vehicle, the baseline bandwidth of the onboard server, the current speed of the target vehicle and the preset bandwidth model of the onboard server; the preset bandwidth model of the onboard server includes the relationship between the bandwidth attenuation coefficient, the baseline bandwidth, the current speed and the current bandwidth.

[0014] In this embodiment, the preset bandwidth model is a pre-built model. Optionally, the preset bandwidth model for the vehicle-mounted server is B = B0 × α × f(v), where B is the current bandwidth of the vehicle-mounted server, B0 is the baseline bandwidth of the vehicle-mounted server, α is the bandwidth attenuation coefficient, v is the current speed of the target vehicle, and f( ) is the vehicle speed attenuation function. Based on this, this embodiment can obtain a relatively accurate current bandwidth of the vehicle-mounted server.

[0015] Among them, B0 is the theoretical maximum bandwidth of the vehicle-mounted server, which is determined by the hardware specifications of the communication module corresponding to the vehicle-mounted server.

[0016] Among them, the bandwidth attenuation coefficient α corresponding to the location of the target vehicle is an empirical value or obtained by fitting historical data or determined according to the base station density at the location of the target vehicle, 0≤α≤1. For example, when the base station density at the location of the target vehicle is greater than or equal to the preset base station density, the bandwidth attenuation coefficient corresponding to the location of the target vehicle is determined to be 1; when the base station density at the location of the target vehicle is less than the preset base station density, the bandwidth attenuation coefficient corresponding to the location of the target vehicle is determined to be the ratio of the base station density at the location of the target vehicle to the preset base station density; among them, the preset base station density is an empirical value, which corresponds to the base station density when the bandwidth of the vehicle-mounted server is not attenuated.

[0017] v can be obtained through the vehicle's CAN bus. If v ≤ v0, then f(v) = 1; if v > v0, then f(v) = max(0, 1 - k × (v - v0)), where k is the vehicle speed attenuation coefficient, v0 is the preset speed threshold, and max( ) is the maximum value. For k > 0, 0 ≤ f(v) ≤ 1. Optionally, v0 is determined based on network switching frequency experiments. For example, if experiments show that the base station switching frequency is higher (i.e., greater than a preset frequency) when the vehicle speed is greater than or equal to 80 km / h, then v0 = 80 km / h. Alternatively, if experiments show that the base station switching frequency is higher when the vehicle speed is greater than or equal to 70 km / h, then v0 = 70 km / h. Optionally, k is determined through historical data regression analysis. For example, k = 0.002 means that for every 1 km / h increase in vehicle speed, the bandwidth attenuates by 0.2%.

[0018] S200, obtaining the transmission bandwidth between the on-board server and each external server end according to the current bandwidth of the on-board server and each preset transmission model in the preset transmission model set; the preset transmission model set includes n preset transmission models, n is the number of external server ends, the i-th preset transmission model corresponds to the i-th external server end, and the i-th preset transmission model includes the relationship between the transmission bandwidth between the on-board server and the i-th external server end and the current bandwidth of the on-board server; i=1,2,…,n.

[0019] In this embodiment, Figure 2 As shown, the vehicle-mounted server can send data to any of n external servers, where n ≥ 2. Optionally, the external server is a cloud server, which has stronger computing power, faster computing speed, larger bandwidth, and relatively more stable network quality than the vehicle-mounted server. When the vehicle-mounted server transmits data to the external server, the transmission bandwidth between the vehicle-mounted server and the external server is mainly limited by the current bandwidth of the vehicle-mounted server.

[0020] In this embodiment, the preset transmission model corresponding to each external server is pre-built; optionally, the i-th preset transmission model is y i =a i ×e ri×x ,y i is the transmission bandwidth between the vehicle server and the i-th external server, a i is the base transmission coefficient of the i-th external server, ri is the influence coefficient of the vehicle server bandwidth on the transmission bandwidth between the vehicle server and the i-th external server, and x is the current bandwidth of the vehicle server. iand ri are empirical values ​​or obtained by fitting historical data. Optional fitting methods include first taking the natural logarithm and then using linear regression to solve. Based on this, this embodiment can obtain a relatively accurate transmission bandwidth between the vehicle-mounted server and the i-th external server.

[0021] S300, obtain the predicted result return time of each external server end based on the transmission bandwidth between the vehicle-mounted server and each external server end, the predicted processing time corresponding to each external server end, the queuing time corresponding to each external server end, the amount of data to be processed by the vehicle-mounted server and the amount of data of the processing results.

[0022] Optionally, the prediction result return time of the i-th external server is g i , g i =(p+q) / y i +c i +d i , p and q are the amount of data to be processed by the vehicle server and the amount of data of the processing results, respectively, i is the transmission bandwidth between the vehicle server and the i-th external server, c i is the predicted processing time corresponding to the i-th external server, d i is the queuing time corresponding to the i-th external server. Based on this, this embodiment can obtain a relatively accurate prediction result return time of the i-th external server.

[0023] Among them, the amount of data to be processed by the on-board server is known, and the amount of data of the processing results can also be estimated in advance. For example, a conversion relationship between the amount of data to be processed and the amount of data of the processing results corresponding to each data type corresponding to the i-th external server end is established in advance. When the data type to be processed and the amount of data to be processed by the on-board server are known, the amount of data of the processing results corresponding to the i-th external server end can be obtained based on the amount of data to be processed and the conversion relationship corresponding to the i-th external server end.

[0024] Among them, the predicted processing time corresponding to the i-th external server end is the ratio of the amount of data to be processed by the on-board server to the processing speed of the i-th external server end in processing the data of the on-board server. The processing speed of the i-th external server end in processing the data of the on-board server is known, for example, 200MB / s; then, when the amount of data to be processed by the on-board server is also known, the predicted processing time corresponding to the i-th external server end can also be obtained.

[0025] Among them, the queuing time corresponding to the i-th external server is affected by the amount of tasks that the i-th external server needs to process before processing the amount of data to be processed by the on-board server. Optionally, the queuing time corresponding to the i-th external server is the ratio of the queue length of the current task queue of the i-th external server (for example, the queue length is 2 tasks) to the average processing speed of the i-th external server (for example, processing 3 tasks per second).

[0026] S400: Determine the external server with the shortest prediction result return time as the external server currently matching the vehicle-mounted server, and send the data to be processed by the vehicle-mounted server to the external server with the shortest prediction result return time for processing.

[0027] In this embodiment, after the on-board server receives the processing result returned by the external server, it performs the next action based on the processing result. For example, if the processing result returned by the external server is a reminder to slow down, the on-board server executes the deceleration instruction; if the processing result returned by the external server is a reminder to turn left, the on-board server executes the turn left instruction.

[0028] In this embodiment, when the target vehicle is in a state of potential safety hazard, the predicted result return time of each external server terminal is obtained, and the external server terminal with the shortest predicted result return time is determined as the external server terminal currently matched with the vehicle-mounted server. The vehicle-mounted server then sends the data to be processed to the external server terminal with the shortest predicted result return time for processing. This facilitates the vehicle-mounted server to quickly receive the processing results and then quickly execute the next action, thereby improving the safety of the target vehicle. The current bandwidth of the vehicle-mounted server is also determined based on the bandwidth attenuation coefficient corresponding to the target vehicle's location and the current speed of the target vehicle. The transmission bandwidth between the vehicle-mounted server and each external server terminal is also obtained based on different preset transmission models corresponding to each external server terminal. This makes the predicted result return time obtained for each external server terminal more accurate, ensuring that the vehicle-mounted server can send the data to be processed to the external server that can receive the processing results the fastest, thereby ensuring the safety of the target vehicle.

[0029] Optionally, before S100, it also includes: judging whether the target vehicle is in a safety hazard state based on the first type of sensor of the target vehicle; the data to be processed by the onboard server is the data of the second type of sensor of the target vehicle; the first type of sensor is a sensor whose collected data is processed by the onboard server; the second type of sensor is a sensor whose collected data is processed by an external server.

[0030] The first type of sensor corresponds to a smaller amount of data and has a simpler judgment logic. For example, for first-type sensors such as steering wheel torque sensors and temperature sensors, if the torque detected by the steering wheel torque sensor remains at zero, the driver is deemed to have taken their hands off the wheel and the target vehicle is deemed to be in a safety hazard state. If the ambient temperature detected by the temperature sensor is greater than a preset temperature value, the target vehicle is deemed to be in a safety hazard state. In contrast, the second type of sensor corresponds to a larger amount of data and has a more complex judgment logic. For example, second-type sensors such as cameras and lidars require a large amount of data processing for the camera's raw video stream. Obstacle type identification based on this raw video stream requires a deep learning model, which is difficult to accomplish on an onboard server. The lidar's point cloud data also requires a large amount of data processing, and constructing a high-precision three-dimensional structure of the road ahead based on point cloud data is also difficult to accomplish on an onboard server.

[0031] As an optional specific implementation, S100 also includes: in response to the target vehicle currently being in a safe state, determining the comprehensive loss of each external server end based on the prediction result return time and processing accuracy of each external server end, and sending the data to be processed by the on-board server to the external server end with the smallest comprehensive loss for processing.

[0032] In this embodiment, the process of obtaining the prediction result return time of each external server end when the target vehicle is currently in a safe state is the same as the above-mentioned process of obtaining the prediction result return time of each external server end when the target vehicle is currently in a safety hazard state, and will not be repeated here.

[0033] Optionally, the comprehensive loss of the i-th external server is L i , L i =w1×t' i +w2×(1-u i ), w1 and w2 are the weights corresponding to time and accuracy, t' i is the value after normalizing the return time of the prediction result of the i-th external server, u i is the processing accuracy of the i-th external server, w1 and w2 are both greater than 0 and less than 1, and the sum of w1 and w2 is 1, 0 i <1. The processing accuracy of the i-th external server is known in advance. Those skilled in the art will appreciate that the normalization process is conventional and will not be described in detail here. Optionally, w1 and w2 are empirical values, for example, w1 = w2 = 0.5.

[0034] ​As another optional specific implementation, S100 also includes: in response to the target vehicle currently being in a safe state, sending the data to be processed by the on-board server to the external server end with the highest processing accuracy for processing. Based on this, the present embodiment can obtain more accurate processing results. Optionally, sending the data to be processed by the on-board server to the external server end with the highest processing accuracy for processing includes: obtaining the sending time of each feasible sending path, and sending the data to be processed by the on-board server to the external server end with the highest processing accuracy for processing according to the feasible sending path with the shortest sending time; any feasible sending path can realize the sending of the data to be processed by the on-board server to the external server end with the highest processing accuracy. Based on this, the present embodiment can take into account the timeliness of sending data while obtaining more accurate processing results. Optionally, the processing result is directly replied to the on-board server through the external server end with the highest processing accuracy.

[0035] Optionally, based on the exhaustive method, all feasible sending paths that can realize sending the data to be processed by the vehicle server to the external server end with the highest processing accuracy are obtained. In view of the fact that the delay of the multi-level transit path is often high, this embodiment only considers the single-hop transit path. For example, there are 4 external server ends, namely the first external server end, the second external server end, the third external server end and the fourth external server end; among them, the processing accuracy of the third external server end is the highest, and the feasible sending paths include: the vehicle server to the third external server end; the vehicle server to the first external server end, and then to the third external server end; the vehicle server to the second external server end, and then to the third external server end; the vehicle server to the fourth external server end, and then to the third external server end.

[0036] In this embodiment, the sending time of any feasible sending path is the sum of the sending times of each segment of the feasible sending path. For example, if a feasible sending path is from the vehicle-mounted server to the second external server end and then to the third external server end, then the sending time of the feasible sending path is the sum of the sending time from the vehicle-mounted server to the second external server end and the sending time from the second external server end to the third external server end. The sending time from the vehicle-mounted server to the second external server end is the ratio of the amount of data to be processed by the vehicle-mounted server to the transmission bandwidth between the vehicle-mounted server and the second external server end. The sending time from the second external server end to the third external server end is the ratio of the amount of data to be processed by the vehicle-mounted server to the transmission bandwidth between the second external server end and the third external server end. Example

[0037] This embodiment provides a multi-server matching device for an in-vehicle server, the device including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the following steps are implemented: S100, in response to the target vehicle currently being in a safety hazard state, the current bandwidth of the onboard server is determined based on the bandwidth attenuation coefficient corresponding to the location of the target vehicle, the baseline bandwidth of the onboard server, the current speed of the target vehicle and the preset bandwidth model of the onboard server; the preset bandwidth model of the onboard server includes the relationship between the bandwidth attenuation coefficient, the baseline bandwidth, the current speed and the current bandwidth.

[0038] S200, obtaining the transmission bandwidth between the on-board server and each external server end according to the current bandwidth of the on-board server and each preset transmission model in the preset transmission model set; the preset transmission model set includes n preset transmission models, n is the number of external server ends, the i-th preset transmission model corresponds to the i-th external server end, and the i-th preset transmission model includes the relationship between the transmission bandwidth between the on-board server and the i-th external server end and the current bandwidth of the on-board server; i=1,2,…,n.

[0039] S300, obtain the predicted result return time of each external server end based on the transmission bandwidth between the vehicle-mounted server and each external server end, the predicted processing time corresponding to each external server end, the queuing time corresponding to each external server end, the amount of data to be processed by the vehicle-mounted server and the amount of data of the processing results.

[0040] S400: Determine the external server with the shortest prediction result return time as the external server currently matching the vehicle-mounted server, and send the data to be processed by the vehicle-mounted server to the external server with the shortest prediction result return time for processing.

[0041] Although some specific embodiments of the present invention have been described in detail by way of example, it should be understood by those skilled in the art that the above examples are for illustration only and are not intended to limit the scope of the present invention. It should also be understood by those skilled in the art that various modifications may be made to the embodiments without departing from the scope and spirit of the present invention. The scope of the present invention is defined by the appended claims.

Claims

1. A multi-server matching method for an in-vehicle server, characterized in that: The following steps are involved: S100, in response to the target vehicle currently being in a safety hazard state, determining a current bandwidth of the vehicle-mounted server based on a bandwidth attenuation coefficient corresponding to the location of the target vehicle, a baseline bandwidth of the vehicle-mounted server, a current speed of the target vehicle, and a preset bandwidth model of the vehicle-mounted server; The preset bandwidth model of the vehicle server includes the relationship between the bandwidth attenuation coefficient, the baseline bandwidth, the current speed and the current bandwidth; S200, obtaining a transmission bandwidth between the in-vehicle server and each external server end according to the current bandwidth of the in-vehicle server and each preset transmission model of the preset transmission model set; The preset transmission model set includes n preset transmission models, where n is the number of external server terminals, the i-th preset transmission model corresponds to the i-th external server terminal, and the i-th preset transmission model includes a relationship between a transmission bandwidth between the vehicle-mounted server and the i-th external server terminal and a current bandwidth of the vehicle-mounted server; i=1, 2, ..., n; S300, obtaining a prediction result return time for each external server based on the transmission bandwidth between the onboard server and each external server, the prediction processing time corresponding to each external server, the queuing time corresponding to each external server, the amount of data to be processed by the onboard server, and the amount of data of the processing results; S400: Determine the external server with the shortest prediction result return time as the external server currently matching the vehicle-mounted server, and send the data to be processed by the vehicle-mounted server to the external server with the shortest prediction result return time for processing.

2. The multi-server matching method of the vehicle-mounted server according to claim 1, characterized in that: The preset bandwidth model of the vehicle server is B=B0×α×f(v), where B is the current bandwidth of the vehicle server, B0 is the baseline bandwidth of the vehicle server, α is the bandwidth attenuation coefficient, v is the current speed of the target vehicle, and f( ) is the vehicle speed attenuation function.

3. The multi-server matching method of the vehicle-mounted server according to claim 2, characterized in that: If v≤v0, then f(v)=1; if v>v0, then f(v)=max(0,1-k×(v-v0)), where k is the vehicle speed attenuation coefficient, v0 is the preset speed threshold, and max( ) is the maximum value.

4. The multi-server matching method of the vehicle-mounted server according to claim 1, characterized in that: The i-th preset transmission model is y i =a i ×e ri×x ,y i is the transmission bandwidth between the vehicle server and the i-th external server, a i is the benchmark transmission coefficient of the i-th external server, ri is the influence coefficient of the vehicle server bandwidth on the transmission bandwidth between the vehicle server and the i-th external server, and x is the current bandwidth of the vehicle server.

5. The multi-server matching method of the vehicle-mounted server according to claim 1, characterized in that: The prediction result return time of the i-th external server is g i , g i =(p+q) / y i +c i +d i , p and q are the amount of data to be processed by the vehicle server and the amount of data of the processing results, respectively, i is the transmission bandwidth between the vehicle server and the i-th external server, c i is the predicted processing time corresponding to the i-th external server, d i is the queuing time corresponding to the i-th external server.

6. The multi-server matching method of the vehicle-mounted server according to claim 1, characterized in that: Before S100, it also includes: judging whether the target vehicle is in a safety hazard state based on the first type of sensor of the target vehicle; the data to be processed by the on-board server is the data of the second type of sensor of the target vehicle; the first type of sensor is a sensor whose collected data is processed by the on-board server; the second type of sensor is a sensor whose collected data is processed by an external server.

7. The multi-server matching method of the vehicle-mounted server according to claim 6, characterized in that: S100 also includes: in response to the target vehicle currently being in a safe state, determining the comprehensive loss of each external server end according to the prediction result return time and processing accuracy of each external server end, and sending the data to be processed by the on-board server to the external server end with the smallest comprehensive loss for processing.

8. The multi-server matching method of the vehicle-mounted server according to claim 6, characterized in that: S100 also includes: in response to the target vehicle currently being in a safe state, sending the data to be processed by the vehicle-mounted server to an external server with the highest processing accuracy for processing.

9. The multi-server matching method of the vehicle-mounted server according to claim 8, characterized in that: Sending the data to be processed by the on-board server to the external server with the highest processing accuracy for processing includes: obtaining the sending time of each feasible sending path, and sending the data to be processed by the on-board server to the external server with the highest processing accuracy for processing according to the feasible sending path with the shortest sending time; any feasible sending path can realize the sending of the data to be processed by the on-board server to the external server with the highest processing accuracy.

10. A multi-server matching device for an in-vehicle server, the device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the multi-server matching method for the vehicle-mounted server according to any one of claims 1 to 9 is implemented.

Images