Task scheduling method based on depth reinforcement learning under vehicle network environment
A technology of reinforcement learning and task scheduling, applied in neural learning methods, biological neural network models, program startup/switching, etc., can solve problems such as tasks that cannot be completed in time
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Publication Date
- 2017-09-08
Smart Images

Figure 1 
Figure 2 
Figure 3
Abstract
Description
technical field
[0001] The invention belongs to the field of computer mobile cloud computing, and in particular relates to a task scheduling method based on deep reinforcement learning in a vehicle network environment. Background technique
[0002] Mobile cloud computing (Mobile Computing) is a new cloud computing model that has emerged with the rapid development of mobile terminals in recent years. It provides abundant computing resources for mobile end users and cloud service providers. The mobile terminal can offload tasks to the cloud, and the cloud returns the calculation result to the mobile terminal to overcome the problem of limited computing power of the mobile terminal and reduce the power consumption of the mobile terminal.
[0003] As a typical case of self-organizing network, vehicular ad-hoc network (VANET) can share data and offload tasks through vehicle-to-vehicle communication (V2V) and vehicle-to-infrastructure communication (V2I). With the development of...
Examples
Embodiment
[0100] In this embodiment, a certain area of city A is used for experiments.
[0101] For this area, there are 10 roadside units, count the number of vehicles in each roadside unit within a certain period of time, unit (vehicle) {Q 1 ,Q 2 ,...Q 10}. Get the task queue length {L of each roadside unit 1 , L 2 ,... L 10}.
[0102] Secondly, initialize the neural network for task assignment as an input layer of 20 neurons, a first hidden layer of 7 neurons, a second hidden layer of seven neurons, and an output layer of 10 neurons .
[0103] Again, warm up the neural network, and record the response time and environment variables of the tasks within a period of time according to the strategy of random assignment.
[0104] Then, the profit value of each strategy is calculated according to the response time, and the profit value is standardized in order to clarify whether the strategy is good or bad.
[0105] Next, the neural network is updated based on the BP algorithm us...