L
Admin
[Technical article, rfforum Editorial Team]
Link budgets often freeze a path-loss exponent n (FSPL n=2, dense urban sometimes 3–4) with no site data behind it. A short drive test with a known CW or pilot and a logged RSSI can estimate n for that environment in an afternoon.
1. Model
Write PL(d) ≈ PL(d0) + 10 n log10(d/d0) + Xσ. In dB-space, RSSI ≈ Ptx + Gt + Gr − PL(d0) − 10 n log10(d/d0) + noise. For fixed antennas and power, a linear fit of RSSI versus log10(d) has slope −10 n.
2. Data hygiene
- Use 2D or 3D distance consistently; for rooftop-to-street include height or you will bias n.
- Exclude near-field / break-point distances (d < d0, often 1–100 m depending on band and clutter).
- Median or mean in log distance bins to reduce fast fading; do not fit raw multipath speckles.
- Record frequency, antenna heights, and whether the path is LOS, light NLOS or hard NLOS — fit them separately if you can.
3. Fitting
Ordinary least squares on binned medians is enough for planning. Report n and the shadowing σ (rms residual). A single n for mixed LOS+NLOS is usually a fiction; two slopes with a break point match cities better.
4. What good numbers look like
Suburban LOS cellular/UHF often lands near 2.2–2.8; dense NLOS indoor corridors can exceed 4. If your fit gives n < 2 in clear LOS, check calibration, antenna pattern toward the route, or GPS distance errors.
5. Using n in the budget
Carry σ into fade margin (e.g. for a target outage). Do not publish a link budget with n from a textbook and σ = 0.
Discussion: what n and σ did you last measure on a sub-6 GHz outdoor route, and did a single slope suffice or did you need a break point?
— rfforum 编辑部 / Editorial Team
Link budgets often freeze a path-loss exponent n (FSPL n=2, dense urban sometimes 3–4) with no site data behind it. A short drive test with a known CW or pilot and a logged RSSI can estimate n for that environment in an afternoon.
1. Model
Write PL(d) ≈ PL(d0) + 10 n log10(d/d0) + Xσ. In dB-space, RSSI ≈ Ptx + Gt + Gr − PL(d0) − 10 n log10(d/d0) + noise. For fixed antennas and power, a linear fit of RSSI versus log10(d) has slope −10 n.
2. Data hygiene
- Use 2D or 3D distance consistently; for rooftop-to-street include height or you will bias n.
- Exclude near-field / break-point distances (d < d0, often 1–100 m depending on band and clutter).
- Median or mean in log distance bins to reduce fast fading; do not fit raw multipath speckles.
- Record frequency, antenna heights, and whether the path is LOS, light NLOS or hard NLOS — fit them separately if you can.
3. Fitting
Ordinary least squares on binned medians is enough for planning. Report n and the shadowing σ (rms residual). A single n for mixed LOS+NLOS is usually a fiction; two slopes with a break point match cities better.
4. What good numbers look like
Suburban LOS cellular/UHF often lands near 2.2–2.8; dense NLOS indoor corridors can exceed 4. If your fit gives n < 2 in clear LOS, check calibration, antenna pattern toward the route, or GPS distance errors.
5. Using n in the budget
Carry σ into fade margin (e.g. for a target outage). Do not publish a link budget with n from a textbook and σ = 0.
Discussion: what n and σ did you last measure on a sub-6 GHz outdoor route, and did a single slope suffice or did you need a break point?
— rfforum 编辑部 / Editorial Team
