Generate synthetic data that’s as good, or even better than the data you have. Or don’t have.
Machine Learning: Real vs. synthetic data
Machine learning algorithms need ever-increasing amounts of data for training and testing.
The problems with real data are that it:
- Can be difficult and expensive to collect. This is particularly true for data that is complex or specialized
- Is time-consuming to label. This process also requires expert knowledge which comes at a high cost.
- May contain sensitive or confidential information
- Is often unbalanced; it may not contain enough examples of certain classes or phenomena
On the other hand, synthetic data can be:
- Generated at very low cost and in vast quantities
- Perfectly labeled, and hence directly used to train neural nets
- Generated to represent a wide variety of scenarios and edge cases, which can help to improve the robustness and generalizability of machine learning models.
- Created to be free of sensitive or confidential information
NetSim Network Simulator
NetSim is used for modeling, simulating, and analyzing a wide range of communication networks. It covers a wide range of technologies (https://tetcos.com/technology-libraries.html), networking scenarios, and edge cases. Shipped along with our automated utilities (https://tetcos.com/netsim-utilities.html) for scenario generation and execution can generate large amounts of data in the form of CSV files.
NetSim Simulation Data
The output results and data files generated by NetSim include:
- Performance Metrics
- Instantaneous and average throughputs for each link and each application
- Buffer occupancy vs. time at source and intermediate devices
- TCP congestion window vs. Time at End each UE and remote Server
- Packet trace
- 30+ parameters for every packet as it flows through the network. These include arrival times, queuing times, departure times, payload, overhead, errors, collisions, etc
- Radio measurements
- SINR, Pathloss, Shadowing, Fast fading, LOS/NLOS states, O2I Loss, MCS, CQI, BS-UE distances, UE-gNB association.
- Radio resource allocation
- Buffer fill (queue size), scheduling metric, PRBs allocated
Users regularly generate data files that reach up to 10 million rows per simulation. Special cases touch 100 million rows.
Our customers: NetSim is used by 500+ organizations across 25+ countries. Top-tier cellular service providers (CSPs) and 5G equipment manufacturers use NetSim to generate synthetic data to train their machine-learning algorithms.