Unlicensed coexistence networks and spectrum sharing are two relatively new technological paradigms in cellular technology. These wireless systems are standardized and adopted to help cellular operators meet the ever-increasing mobile data demand by efficient utilization of unlicensed bands. However, several incumbents are already operational in these frequencies such as military, radar, and navy systems rendering the wireless environment extremely dynamic and unpredictable. Consequently, greater probabilities of transmission conflicts and differing quality-of-service (QoS) requirements present new challenges in harmonious coexistence and spectrum sharing. Compared to conventional optimization models, machine learning-based data-driven solutions are more suitable for network analysis and performance optimization.

However, a machine learning model's performance is determined by the data it is fed. Thus, machine-learning-driven solutions for unlicensed cellular networks in the New Radio in Unlicensed (NR-U) in 6GHz, will need cellular operator data from current Licensed Assisted Access (LAA) deployments. Unfortunately, due to expensive network monitoring applications and limited deployment of unlicensed networks access to network data is limited. This dataset bridges this gap by making LAA network data from three major cellular operators in Chicago available to the wider research community.


The dataset includes a total of fifteen network features, that represent twelve important unlicensed network parameters that include SINR, RB, and Throughput. It comprises 9676 samples from the LAA deployments of the three networks. The

empty fields in the dataset indicate that the NSG screens were empty for these fields. Further, there is a one-to-one correspondence between all CWD0 and CWD1 variables. 

Funding Agency: 
Japan Science and Technology Agency (JST) and National Science Foundation (NSF)
Grant Number: 
JST CREST Grant Number JPMJCR21M5 and NSF Grant No. CNS-1618920