Edge Intelligence in Softwarized 6G.md · GitHub
https://share.summari.com/edge-intelligence-in-softwarized-6gmd-github?utm_source=Chrome
- The 6G vision is envisaged to enable agile network expansion and rapid deployment of new on-demand microservices (e.g., visibility services for data traffic management, mobile edge computing services) closer to the network's edge IoT devices.
- providing one of the critical features of network visibility services, i.e., data flow prediction in the network, is challenging at the edge devices within a dynamic cloud-native environment as the traffic flow characteristics are random and sporadic
- A novel edge-native framework to provide an intelligent prognosis technique called deep learning is proposed which accurately predicts the statistical characteristics of data traffic and verifies the trained model against the ground truth observations
- During the network operations, the data traffic flow having a timeseries (TS) data nature behaves non-linearly with aperiodic characteristics in an increasingly dynamic and complex network environment
The current networking infrastructure needs to evolve and adapt cloud-native AI deployment strategies to provide a balanced support for both the AI-based microservices functions and other diverse network functions
One potential solution to approaching the preceding challenges is adopting edge intelligence
A novel idea of an edge intelligence method for analyzing and predicting network data traffic at edge devices according to the emerging 6G vision of softwarized networks
Place theseμ-boxes at the multi-site edge locations of the OPG, having storage-storage-networking resources to allow IoT-SDN/NFV-Cloud functionalities
Collect the time series raw traffic flow data at the edge of the network, which is sent to the visibility center for storage and processing
Orchestrate the Kubeflow deployment at the orchestration center, used to develop and train the DL model for the prognosis
Train the model on collected TS data and predict the statistical properties of data traffic
Analyze the developed model’s performance in terms of root-mean-square error (RMSE) and coefficient of determination (R^2) metrics
The collected dataset comprised 43000 time-series records with five features representing statistical features of data flow, divided into training (65%) and test/validation (35%).
To train the prediction model, they deployed Kubeflow and trained the LSTM-based encoder-decoder model on Jupyter.
They trained the model on the past 20 hours of observation to predict the next 10 hours of output samples, compared with the groundtruth observations for model validation.