User association joint power distribution strategy based on deep learning in heterogeneous network

Abstract

The invention discloses a user association joint power distribution strategy based on deep learning in a heterogeneous network, which belongs to the technical field of communication systems, and comprises the following steps: step 1, communication modeling: establishing a heterogeneous network model; 2, collecting a data set; 3, constructing a neural network; 4, training the neural network, and determining an optimization algorithm as Adam; and 5, saving the network when all iterations are satisfied. According to the user association joint power distribution strategy based on deep learning in the heterogeneous network, in the heterogeneous network, the neural network is used for jointly optimizing user association and power distribution in a supervised learning mode, high fitting of a traditional algorithm is achieved, meanwhile, low calculation complexity is guaranteed, and real-time and reliable online decisions are provided.

Description

technical field [0001] The invention belongs to the technical field of communication systems, and in particular relates to a user-associated joint power allocation strategy based on deep learning in a heterogeneous network. Background technique [0002] With the rapid development of the mobile Internet, the ever-increasing data services have gradually increased the requirements for the capacity and speed of the cellular network. At present, the architecture of the cellular network is gradually evolving towards heterogeneity. Small base stations with low energy consumption form a heterogeneous network and provide users with more spectrum resources to improve user rate performance. In a traditional cellular network, end users are associated with macro base stations, but the number of macro base stations is limited. End users usually measure the signal power of each macro base station and choose the strongest base station to associate with, which will inevitably lead to low pow...

AI Technical Summary

Problems solved by technology

In a traditional cellular network, end users are associated with macro base stations, but the number of macro base stations is limited. End users usually measure the signal power of each macro base station and choose the strongest base station to associate with, which will inevitably lead to low power in the coverage area. The load difference between the small base station and the macro base station is too large, so it is necessary to actively associate the terminal equipment with the small base station to make more effective use of the wireless resources of the small base station. A simple solution is to expand the range of the cell and artificially enlarge it from The signal power of the small base station makes more users choose to associate with the small base station, but the power amplification factor is difficult to determine the optimal value

It is also a solution to convert the relaxed constraints into a convex optimization problem and use the Lagrange dual solution. In addition, it can also be solved by solving the matching problem of the bipartite graph to obtain the user association strategy, but the above-mentioned solution has high computational complexity. In the ever-changing communication field, it cannot be used for real-time decision-making

[0003] After the user association strategy is determined, how to allocate the downlink transmission power of a single base station is also a point of concern. The traditional power allocation algorithm is an iterative water filling algorithm. Power, users with poor channel quality use low power to maximize the utility function of a single base station, but the algorithm has poor convergence performance, including low convergence, and slow convergence will lead to high computational complexity. also limits its scope of application

Images