Wireless networks can now pick antenna ports faster without burning through computing power

A new method lets wireless systems with multiple antenna ports choose which ones to use for transmission without the expensive computation that usually comes with that choice. Instead of calculating every possible combination, a greedy algorithm handles most of the work, then a trained neural network runs the refinements — cutting the compute cost while keeping the signal quality.

Wireless systems have always faced a hard trade-off: optimize antenna selection perfectly and you burn CPU time, or use shortcuts and lose signal quality. This work breaks that trade-off by combining a simple greedy approach with a neural network trained to mimic the expensive method. What changes in practice is that network operators can now handle multi-antenna systems without overprovisioning hardware just to run the port-selection algorithm — that's a cost reduction that scales across every deployed system doing this calculation.