...
- Once the all the training data are acquired, the training module performs preprocessing to convert the data from json JSON format into Dictionary format
- In Preprocessing the training Data is normalised normalized and Transformed.
- The preprocessed data (PM data Parameters data content) for all slices and cells which are in time series are converted into forcast forecast series.
- The forecast series is cells PM data parameters for current time instance, Plus the previous past or last four time instance data.
- Similar forecast series data is prepared for all the cells in the slice.
- Data are split into training and validation datasets
- The expected results for each time instance cell data are synthetically generated based on rule conditions and the cell intends.
- The training module perform Training against an LSTM model.
- The Training model file are archived.
- Sample training data referred at following source code location components/ml-prediction-ms/train/ExampleSample_train_data_s1.xlsx
- P.S: This file is only for reference, on how the training data are prepared. In actual training process, the data are fetched RANSim topics.
- Following wiki attached excel file contains sample cell data from a slice. Here each of the cell's timeseries data will appended from first cells to the last cell along with respective its cells time series to that slice. or the cells are append successively within the slice. This action in training module results in large training data to train the model.
Spreadsheets containerType PAGE documentId ExampleSample_train_data_s1_smaller.xlsx_0.ssc containerId 117743292 documentTitle title ExampleSample_train_data_s1_smaller.xlsx - Here on each cell parameters like PDUSessionSetupReq, PDUSessionSetupSucc & PDUSessionSetupFail are obtained from RANSim or PM Data Topics.
- Parameter maxNumberOfConns_configured also obtained from configDB Topic.
- The prediction parameter for each timeseries are computed for each cells independently with respect to previous time series data.
- By change in current ratio on PDUSessionSetupFail verses PDUSessionSetupReq minus the respective cells previous time instance PDUSessionSetupFail verses PDUSessionSetupReq, this results in computation on +/- change percentage that needs to applied against the config data.
- The training module perform Training against an LSTM model.
- The Training model file are archived.
- The The results are manually verified .
...
- Following are the processes involved in the Prediction the model.
- The data from above PM DATA topics are used to acquire the current time instance and also the previous last four time instance data for all cells and slice.
- The collected data are parsed, normalisednormalized, transformed and converted into forcase forecast series, the steps followed here are almost the same to the input preprocessing steps used in Training module.
- The preprocess forecast series along with the intend are fed to the ML model to perform prediction for the resource allocation for each cells.
- The predicted results are are further optimised processed in the post processing were they are optimized by the 25% +/- on w.r.t cell config and 10% +/- on /- w.r.t aggregation on overall resources available at slice config thresholds.
- Recommendation for each cell can be maximum of 5120 and the 25% +/-. The aggregation of maximumNoOfConnections recommended for a slice (sNSSAI) can be upto 110% (10 % buffer) of the intent.
- Now predict recommend suggestions for the cell level configurations (maxNumberOfConns) for a slice. A new micro service will be introduced to do the data processing and prediction for the dynamic data.each slice and cells are updated via DMAP POLICY UPDATE Topic.
Suggested Configuration from the ML MS DMAP POLICY UPDATE Topic should be in the form of below format were the predictedMaxNumberofConns values are updated for all cells in slices.
Expand | ||
---|---|---|
| ||
{ |
...