Generate and Test DataSet of the Network Slice Request
tags: docs
Network Slice
Introduction
Create DataSet by yaml file and Test algorithm performance
Git Clone
$ git clone https://github.com/p76081158/5g-nsmf.git
Create Dir of DataSet in slice-requests
$ cd 5g-nsmf/slice-requests
$ mkdir <DataSet Name>
DataSet is generated by yaml
Create yaml file
$ cd <DataSet Name>
$ nano DataSet-parameter.yaml
DataSet-parameter.yaml
datasetInfo:
name: DataSet-Example # name of the dataset
ngciList: # tenants list
- 466-01-000000010
- 466-11-000000010
- 466-93-000000010
sliceNum: 10 # slice number of each tenants
extraRequest: 1 # each timewindow has n network slice requests ( n = tenant num + extraRequest, e.g. n = 3 + 1 )
testNum: 5 # how many test case in this dataset
resource:
cpu:
limit: 1000 # cpu request limit (1000 = 1 cpu core)
lambda: 500 # generate slice reqeusted cpu by poisson distribution
discount: 0.5 # forecasting discount
bandwidth:
limit: 10 # bandwidth reqeust limit
lambda: 5 # generate slice reqeusted cpu by poisson distribution
discount: 0.5 # forecasting discount
duration: 300 # duration of slice reqeust
random: false # slice request duration is fixed or random (poison lambda = duration)
target: cpu # target of resource for slice bin packing algorithm
sort: true # sorting network slice requests in each timewindow before bin packing
concat: false # use concat algorithm if not finding any tree node for slice reqeust
timewindow: 600 # duration of each timewindow
forecastingTime: 0 # from which timewindow start use forecasted data (0 == nerver, 1 == from timewindow-1)
forecastBlockSize: 100 # forecasted block size of network slice request, e.g. each network slice reqeust's duration is 300, so each slice requests will be divided into two sub slice requests (300/150=2)
regenerate: true # regenerate DataSet again (false for DataSet already exist)
Usage
$ cd 5g-nsmf/nsrtester
# ./nsrtester <dataset name>
$ ./nsrtester DataSet-Example