Good RSRP, Bad SINR: Splitting Interference from Overshoot
An LTE 1800 MHz three-sector macro site serves a small market town. Sector 2 (azimuth 120°) faces a low-rise housing estate and is the worst performer in the weekly report. Over three weeks sector 2 shows handover success of 91.8%, against 99.1% and 99.3% for sectors 1 and 3, and uplink RSSI per PRB has risen from about −105 dBm to −93 dBm, the rise being largest between 19:00 and 23:00. Drive tests in the estate give good RSRP (−85 to −95 dBm) but SINR that falls to −2 to 3 dB on many samples. VSWR at the sector 2 antenna is 1.3 and there are no hardware alarms. The timing advance distribution shows 18% of samples in the final TA bin, against 4% for the other sectors. A three-storey shopping complex was completed 400 m from the site three weeks ago, and a 10 kW FM broadcast mast stands 250 m away on the same rooftop.
1. Using the data given, separate the interference problem from the coverage and handover problem.
Read the worked answer
Good RSRP with poor SINR, alongside a 12 dB rise in uplink RSSI per PRB, is the signature of uplink interference rather than weak coverage. If coverage were the problem, RSRP would be poor at the same test points where SINR collapses.
The shape of the RSSI rise carries information. A rise roughly equal across all PRBs points to a wideband source or to receiver desensitisation; a rise concentrated in a few PRBs points to passive intermodulation inside the site's own hardware, or to a leaking repeater.
The timing advance distribution is a separate fault. 18% of samples in the final bin means a large share of users sit at maximum timing advance: sector 2 is overshooting its design footprint, most likely from too little downtilt or too high an antenna. That stretches the handover zone until the neighbour's RSRP is already too weak, which produces handover failures even though RSRP looks healthy at street level.
So there are two independent faults: an interference source, and an overshooting sector. Tuning handover parameters alone would redistribute the failures rather than remove them.
2. Describe a test that shows whether the −93 dBm is entering through the antenna or being generated inside the site, and what each outcome would mean.
Read the worked answer
Put sector 2 into a receive-only state (mute the transmit path or use the site's RX-only monitoring mode) and measure the uplink noise floor at the antenna port with a spectrum analyser, twice: once with the antenna connected, once with a matched load in place of the antenna.
Floor falls to about −105 dBm with the load fitted and stays high with the antenna connected: the energy arrives through the antenna. That means an external emitter, or intermodulation generated outside the cabinet, for example in corroded mast hardware or in re-radiation from the new building's metal cladding.
Floor stays high with the load fitted: the site is generating it internally, through PIM in its own jumpers and connectors, a faulty low-noise amplifier, or a leaking repeater on the same roof.
Given a 10 kW FM mast 250 m away, calculate the third-order products of the FM carriers and see whether they land in the 1800 MHz uplink band, and inspect the site's own rooftop hardware for corrosion and loose connectors. Run this test before changing any parameter, because the answer decides whether the fix is a filter, a jumper, or a conversation with the broadcast operator.
3. Set out the repair plan in priority order and state what you would not do.
Read the worked answer
- Confirm and remove the interference. If the test points outside the site, correct what is inside your control (torque and weatherproof every connector on the sector, replace oxidised jumpers and clamps, fit a filter if the FM products fall in band) and escalate to the broadcast operator with the measured products.
- Correct the overshoot. Increase downtilt and, if needed, lower the antenna so the timing advance distribution returns towards the design footprint, then re-drive-test the estate and confirm SINR improves.
- Re-check neighbour relations and handover thresholds only now. Applied earlier, they merely move failures around an overshooting, noisy cell.
- Verify with a full week of KPIs against sectors 1 and 3, and close the work order only if handover success and RSSI both recover.
What I would not do: raise handover hysteresis or thresholds to suppress the drops, and raise the RSSI alarm threshold so the report looks clean. Both turn a measurable fault into a silent one, and the customer experience in the estate does not change by a single decibel.