CellMapper — Cellular Tower and Signal Map

A phone that reports its own signal strength turns its carrier into a map. CellMapper — Cellular Tower and Signal Map is the result: a crowdsourced survey of cellular infrastructure - filed under Water Utilities, pooled in from wireless telecommunication services, operated by CellMapper Serviced Limited - in which an Android app uploads signal readings (RSRP on LTE, RSSI elsewhere) tagged with GPS positions and cell identifiers, and the backend triangulates approximate tower locations as contributors travel.

What comes back is structural, not statistical. Every tower carries calculated coordinates, a six-way type classification (macrocell, microcell, picocell, DAS, COW - cell on wheels - and decommissioned), per-operator band, frequency and bandwidth detail, LAC/TAC region codes, first-seen and last-updated dates, and a verified flag separating confirmed placements from low-accuracy estimates. Filters extend to LTE carrier aggregation and 5G ENDC capability. Lookups run by location, tower, BSIC/PCI/PSC, cell and LAC/TAC; companion calculators handle ARFCN and eNB ID arithmetic; rendering spans five base maps including OpenStreetMap.

Honesty note: this record scores 6/10 against a 7.81 collection average, largely because field definitions are reconstructed from documented map behavior rather than read off a published schema - which is exactly why this page writes every definition out instead of gesturing at missing documentation. Get a sample of this dataset and read real rows before committing pipeline time.

The underlying record ships no captured extracts, so treat the block below as documented grain - one row per tower, every column traceable to the field dictionary beneath it - with placeholder values standing where real observations would sit:

# grain: one row per tower - documented fields only, placeholder values

tower_lat 51.50742   tower_lon -0.12789   tower_type macrocell    verified TRUE
network_mcc 234      network_mnc 15       technology LTE          bands B1+B3+B7
bandwidth 20 MHz     carrier_aggregation YES                     endc_capable NO
region_lac_tac 21004 first_seen 2019-06-02                       last_updated 2026-08-14

tower_lat 40.71283   tower_lon -74.00605  tower_type microcell    verified FALSE
network_mcc 310      network_mnc 260      technology 5G NR        bands n77
bandwidth 100 MHz    carrier_aggregation NO                      endc_capable YES
region_lac_tac 51227 first_seen 2023-11-19                       last_updated 2026-08-21

Read the anatomy rather than the digits. network_mcc and network_mnc are the join keys - Mobile Country Code plus network code pin each row to exactly one mobile network operator, which turns rival-scoping queries into a filter instead of a project. region_lac_tac groups a tower's cells for clean aggregation. verified is the confidence dial: TRUE marks confirmed placements, FALSE leaves calculated estimates that deserve lighter weight in any model. first_seen date-stamps infrastructure as it appears, so a year-over-year diff of one metro reads as an activation ledger. The spectrum columns - bands, frequency, bandwidth - are the sector detail coverage-area sources never carry.

Six fields anchor the tower grain. One caveat stated plainly: definitions were reconstructed from documented map behavior during the August 2026 research pass - the map exposes filters and flags, not a schema document - so each entry says what the interface proves, nothing more.

• Position: tower lat/lon - calculated by averaging contributor measurement positions, refinable by logged-in contributors. Planning-grade estimate, never survey-grade.
• Classification: tower type - the six-way enum from macrocell down to decommissioned, which lets you exclude temporary COW deployments from permanent-infrastructure counts.
• Spectrum: bands / frequencies / bandwidths - per-sector radio detail, filterable by provider, network and band; a three-sector tower flattens to three spectral records.
• Region: region (LAC/TAC) - the Location Area Code or Tracking Area Code grouping the tower's cells, the natural roll-up key.
• Lifecycle: first seen / last updated - observation start and most recent touch, both exposed as filters and together the basis of any network-evolution cohort.
• Trust: verified flag - confirmed placement versus calculated or low-accuracy estimate; weight accordingly before modeling.

Technology generation (GSM, UMTS, LTE, 5G NR), carrier-aggregation status and 5G ENDC capability ride alongside as documented attributes, so the flat table answers "what did each operator build, on which spectrum, when" without a single join.

• Geography: worldwide, organized by ITU Mobile Country Code with per-country network codes underneath - millions of towers across roughly 200 countries, though no exact total is published anywhere. Density follows contributor travel: national markets run deep, rural patches stay thin, so measure local depth before promising local answers.
• Temporal: rolling crowdsourced observations since launch, with per-tower first-seen and last-updated dates filterable in the interface. That per-row dating converts a static tower list into a time series - watch a carrier's mid-band build-out spread quarter by quarter.
• Grain: individual tower and cell-sector level, the finest structural resolution in the water-utilities pool. Where OpenCelliD answers "where can a handset probably connect" and WiGLE answers "what did devices observe," this feed answers "what steel did operators actually plant, and on which spectrum."

API, files, or your warehouse. Daily, weekly, or hourly.

Take the worldwide tower table once - every MCC in scope, typed columns, tower-type enums decoded, verification flags preserved as booleans - then keep it rotating at whatever cadence your planning cycle expects. You choose the slice: one operator's MNC inside one country, one band across a region, a bounding box around a service territory, macrocells only, or everything carrying a 5G ENDC flag. A sample ships first either way - real rows for your declared scope before any commitment, in the same field shape production deliveries use.

Five jobs the tower grain settles outright:

1. Scoping rival operators - band deployments and new-site activations read straight off MCC/MNC-filtered views, with first-seen dates turning sightings into an activation timeline; see competitor tracking.
2. Smart-meter backhaul and telemetry siting - check whether a proposed collector site will hold a cellular link before trenching for fiber, cross-checked against OpenCelliD coverage radii and WiGLE observation history.
3. Field-workforce coverage expectations - compute signal likelihood across a service territory before dispatching crews, instead of discovering dead zones through missed meter reads.
4. Training coverage models - structural ground truth of where infrastructure actually stands gives signal-prediction models something firmer than drive-test traces alone; see ML model training.
5. Infrastructure reporting and research - spot-check an operator's coverage claims against what its network physically contains, at sector detail that coverage-area maps blur past.

The common thread: whenever the question is sector-level RF structure rather than coverage area, this is the only dataset in the vertical that answers it.

Competitive-intelligence product teams get the sharpest fit - this dataset backs the mined "scope rival operators" job, where band-by-band deployment views and dated activations matter more than raw tower counts. Developers and builders (see developers builders use cases) use it as the verification layer beneath backhaul and telemetry siting, after OpenCelliD narrows the search. Data scientists (see data scientists use cases) find structural ground truth for coverage models, with the edge that tower positions describe infrastructure, not behavior. Journalists, academics and students (see journalists academics use cases) land a modest relevance 1 of 3 - useful for spot-checking coverage anecdotes behind a story, thin for anything longitudinal. Sales-growth teams should pass: steel in the ground says nothing about anyone's budget. Persona fit has edges here, and pretending otherwise wastes a sample request.

Notes worth reading next:

• OpenCelliD — World's Largest Open Database of Cell Towers - the area view to this site view: tens of millions of logical cells across roughly 220 countries with per-cell coordinates and coverage radius. Start there for extent, come here for sector detail.
• WiGLE — Wireless Network Mapping (WiFi + Cell Tower Observations) - roughly 2.0 billion WiFi networks across 27 billion timestamped observations since 2001; the observation history that tests whether mapped structure shows up in measured experience.
• Eurostat Databrowser — ICT and Mobile Internet Use - the demand side: harmonised household adoption statistics that pair with supply-side tower counts when sizing connectivity programs.
• USGS Water Services - Instantaneous, Daily and Statistics REST APIs - the hydrology ledger pooled in from Multi-Utilities, for teams whose day job is water and whose side quest is keeping remote gauges online.

Two glossary notes sharpen the vocabulary: what crowdsourced cell tower data actually contains, and how a cell tower database differs from an observation log. Then go deeper on the ranked best water utilities datasets shortlist or the pooled view on the water utilities data hub.