Abstract
Network Flow Queuing Delay Prediction for City Public Services Based on Long Short-term Memory
Highlights
In computer and telecommunication engineering, there is a queuing delay when the required processing data waits in a series or cascade until it can be executed, which is a very important delay in signal transmission networks
We find that the long shortterm memory (LSTM) framework with three-scale integration has the best prediction accuracy but sacrifices real-time performance owing to large-scale overlapping
We compare the two values of LSTM and LSTM with three-scale integration (LSTM-MERGE) with other representative frameworks,(15) and we record their performances for different prediction scopes, as measured by the root mean square error (RMSE)
Summary
In computer and telecommunication engineering, there is a queuing delay when the required processing data waits in a series or cascade until it can be executed, which is a very important delay in signal transmission networks. As network congestion is inevitable, investigating the possibility of increasing the bandwidth has limited effectiveness in enhancing service performances because it is an uneconomical method. This makes research on delay prediction very important. We use the recurrent neural network (RNN) to propose an improved version of LSTM based on the Spring Cloud system and use this system to predict the network traffic of city public services. We use the RNN-improved multiscale LSTM framework as a specific method and study its actual application, and we compare its prediction results with those of other similar algorithms. The proposed framework can be applied in special cases, for example, a special holiday module can be added for the quantitative analysis of network traffic
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
