Skip to content

MAP_TOWER_FIX.md

Latest commit

 

History

History
157 lines (122 loc) · 5.71 KB

MAP_TOWER_FIX.md

File metadata and controls

157 lines (122 loc) · 5.71 KB
Raw

Cell Tower Location Bug Fix

Problem Summary

Cell towers were being displayed at incorrect physical locations on the map. They were using a simplistic circular distribution around the user's location rather than realistic position estimation based on signal strength.

Root Cause Analysis

Issue 1: Hardcoded Circular Placement

File: MapViewModel.kt (lines 149-153)

The old code was spreading towers in a simple circle:

val angle = (index * 60.0) * (Math.PI / 180.0)
val radius = 0.0004 // About 40 meters (static distance)
val towerLat = latitude + (radius * kotlin.math.cos(angle))
val towerLon = longitude + (radius * kotlin.math.sin(angle))

Problems:

  • Used only the tower index for angle calculation, making placement arbitrary
  • Fixed radius of ~40 meters for all towers regardless of signal strength
  • Ignored the sophisticated CellTowerLocationEstimator utility that was already implemented
  • Towers didn't reflect their actual distance from the user based on signal strength

Issue 2: Unused CellTowerLocationEstimator

File: CellTowerLocationEstimator.kt

The codebase already had a complete location estimation system that:

  • Calculates tower distance using path loss models (free-space, urban, indoor)
  • Uses different reference signals for each network type (2G/3G/4G/5G)
  • Applies network-specific path loss exponents for accurate distance estimation
  • Uses cell ID to determine consistent angular direction (so same tower always appears in same spot)
  • Converts distance to latitude/longitude offsets with proper Earth radius calculations

But it was completely ignored by the MapViewModel.

Solution Implemented

Import CellTowerLocationEstimator

Added import statement to MapViewModel:

import com.stingrayshield.util.CellTowerLocationEstimator

Replace Position Calculation for All Network Types

Changed from hardcoded circular placement to intelligent position estimation for each cell type:

LTE (4G):

val (towerLat, towerLon) = CellTowerLocationEstimator.estimateTowerPosition(
    latitude, longitude, sig.dbm, "LTE", cellId
)

WCDMA (3G):

val (towerLat, towerLon) = CellTowerLocationEstimator.estimateTowerPosition(
    latitude, longitude, sig.dbm, "3G", cellId
)

GSM (2G):

val (towerLat, towerLon) = CellTowerLocationEstimator.estimateTowerPosition(
    latitude, longitude, sig.dbm, "2G", cellId
)

CDMA:

val (towerLat, towerLon) = CellTowerLocationEstimator.estimateTowerPosition(
    latitude, longitude, sig.dbm, "CDMA", id.basestationId
)

5G NR:

val (towerLat, towerLon) = CellTowerLocationEstimator.estimateTowerPosition(
    latitude, longitude, sig.dbm, "5G", nci.toInt()
)

How the Fix Works

Signal-Based Distance Estimation

The estimator uses the path loss formula:

signalStrength = referenceSignal - 10 * n * log10(d/d0)
Solving for d: d = d0 * 10^((referenceSignal - signalStrength) / (10 * n))

Where:

  • referenceSignal: Reference signal at 1 meter (varies by network type)
  • n: Path loss exponent (2.0 for free space, 2.7-3.5 for urban, 3.0-5.0 for indoor)
  • d: Calculated distance in meters
  • Clamped range: 50 meters (min) to 35 km (max) for realistic tower distances

Consistent Angular Direction

val angle = (cellId % 360) * (PI / 180.0)

Uses the cell ID modulo 360 to generate a consistent angle. This means:

  • Same tower ID always appears in the same direction
  • Different towers get different angular positions
  • Much more realistic distribution

Accurate Geographic Conversion

val latOffset = (distance * cos(angle)) / 111320.0
val lngOffset = (distance * sin(angle)) / (111320.0 * cos(userLat * PI / 180.0))

Converts distance in meters to geographic offsets using:

  • 1 degree latitude ≈ 111,320 meters
  • 1 degree longitude ≈ 111,320 * cos(latitude) meters

Network-Specific Parameters

The estimator uses different parameters for each network type to account for their propagation characteristics:

Network Reference Signal @ 1m Path Loss Exponent Frequency Range
5G NR -40 dBm 2.5 Sub-6 GHz / mmWave
4G LTE -45 dBm 3.0 700-2600 MHz
3G UMTS -50 dBm 3.2 850-2100 MHz
2G GSM -55 dBm 3.5 850-1900 MHz
CDMA -50 dBm 3.0 800-1900 MHz

Expected Improvements

✅ Stronger signal towers appear closer to the user's location ✅ Weaker signal towers appear further away ✅ Same tower ID always appears in the same direction ✅ Realistic distances based on cellular propagation models ✅ Network-aware positioning (5G behaves differently than 2G) ✅ Better visual distribution on the map for detection analysis

Testing Recommendations

  1. Move to different locations and verify tower positions adjust accordingly
  2. Compare signal strength with displayed distance (stronger = closer)
  3. Check consistency - same towers should always be in similar positions relative to you
  4. Test different network types (LTE, 5G, etc.) to see different distance characteristics
  5. Verify map patterns - primary/serving towers should typically be closer than neighbor cells

Files Modified

  • app/src/main/java/com/stingrayshield/ui/screens/map/MapViewModel.kt
    • Added import for CellTowerLocationEstimator
    • Updated createCellTower() function for all network types (LTE, WCDMA, GSM, CDMA, NR)

Build Status

✅ Build successful with no compilation errors ✅ All deprecation warnings are pre-existing (unrelated to these changes)