Resource reservation method and system based on flow direction prediction, as well as mobile communication system

Abstract

The invention belongs to the technical field of central resource management and resource consultation, and discloses a resource reservation method and system based on flow direction prediction, as well as a mobile communication system. The resource reservation method comprises the steps of: acquiring historical data of flow direction between cells; analyzing and pre-processing the data; performingdivision and feature extraction for the processed data set; selecting different models to perform training, adjusting parameters to achieve the highest prediction accuracy of the model; performing real-time fusion for prediction results of two models; performing resource reservation according to the prediction result by a network; and determining whether a spacial flow direction operation is required by judging the relationship between the number of the reserved resources and the number of the resources which the cell can provide. The resource reservation method of the invention has the beneficial effects that: real-time changes can be performed according to network characteristics through model fusion; at the same time, the phenomenon of excessive or insufficient resource reservation ina conventional reservation algorithm can be avoided through adoption of a resource reservation algorithm designed in the method, and load balance of the network can be achieved through the spacial flow direction operation.

Description

technical field [0001] The invention belongs to the technical field of central resource management and resource negotiation, and in particular relates to a resource reservation method and system based on flow direction prediction, and a mobile communication system. Background technique [0002] At present, the existing technologies commonly used in the industry are as follows: One is to reserve resources after the mobile node is handed over. The problem with this technology is that the resource reservation process takes a certain amount of time, and user experience cannot be guaranteed for a period of time after the handover. The second is to fully reserve resources before user switching. The problem brought about by this technology is that the resource utilization rate is not high. The third is resource reservation based on prediction. For example, a spectrum pool is used to manage idle spectrum resources, and the bandwidth resources required by users in the next stage are...

AI Technical Summary

Problems solved by technology

[0002] At present, the existing technologies commonly used in the industry are as follows: One is to reserve resources after the mobile node is handed over. The problem with this technology is that the resource reservation process takes a certain amount of time, and the user experience cannot be guaranteed for a period of time after the handover

The second is to fully reserve resources before user switching. The problem brought by this technology is that the resource utilization rate is not high

The third is resource reservation based on prediction. For example, a spectrum pool is used to manage idle spectrum resources, and the bandwidth resources required by users in the next stage are predicted through the transmission rate; however, due to the large delay of the transmission rate prediction model, the bandwidth allocation Real-time performance is difficult to guarantee and can only be applied to small-scale networks; the method based on the Markov chain predicts the load status at the next moment through the Markov transfer function, and this method cannot achieve high prediction accuracy for non-stationary sequences degree, and in the case of big data, it is necessary to establish a multi-dimensional Markov state transition diagram for the number of switching between cells, which cannot achieve the effect of short calculation time and simple model; use the wavelet neural network model to predict each area in the next hour Although it can achieve high prediction accuracy, it cannot make real-time changes according to network characteristics.

[0004] (1) Conventional resource reservation technology cannot achieve high resource utilization

[0005] (2) Although the prediction-based resource reservation technology can improve resource utilization, for a single model, due to the complexity of the network environment, the robustness of the model cannot be guaranteed, so real-time changes cannot be made according to network characteristics

[0006] (3) When the reserved resources exceed the total number of resources that the cell can provide, the existing algorithm allows the handover user to randomly access a nearby neighbor cell in order to prevent the handover user from dropping calls, without taking into account the resource occupancy of the surrounding neighbor cells. It will affect the user experience of the neighboring area

If the model is not properly selected, there will be a large error in the prediction results, which will cause too much or too little resource reservation when the network reserves resources based on the prediction results, which will affect resource utilization.

Second, because the network environment is relatively complex, how to make the model correct in real time according to the characteristics of the network is also the difficulty of the above-mentioned technical problems

Images