the data that go through the Ater PS and Abis are coming from the RLC layer. It includes RLC signalling,, retransmission, etc, and also RLC headers. The RLC blocks carry small amount of data, so the ration header vs data is not as efficient as LLC.
On the Gb interface, the only data are user data + network signaling (GMM / SM). The headers are small, the frames are big. The ratio header vs data is more favourable.
On top of that, the way the Ater and Abis resources are managed is not as efficient as the way the Gb interface is managed (on Ater/Abis : E1 = circuit switch, on Gb : Frame Relay or IP = packet switch).
So you understand why Ater PS congestion can happen easily, while Gb congestion doesn’t.
On top of that, NPO does not provide Gb congestion indicators. The closest thing you get to Gb congestion is DL LLC frames rejected due to congestion, but that is congestion in the MFS, not congestion in the Gb links.
To know Gb congestion, you must know the Gb usage (LLC throughput) compared to the Gb capacity.