RAN Platform
Plan, ingest, analyse, detect, reason, test, change.
Create, manage and optimise multi-vendor, multi-technology RAN networks with ARANO's agentic platform. From network planning to autonomous optimisation.
01
Network Design
Network planning from blank map to fully parameterised site and cell design, greenfield or brownfield, single or multi-technology, driven by real population density data.
Cell Planning
Automated site selection and cell placement from CSV imports or drawn geospatial polygons.
RF Configuration
Antenna tilt, azimuth, power, and beamwidth driven by population density.
Neighbour Planning
Automatic neighbour relation generation based on distance, overlap, and handover analysis.
PCI Assignment
Collision-free PCI plans using collision, confusion, mod-3, and mod-30 conflict detection.
Capacity Dimensioning
PRB and throughput modelling based on traffic forecasts and frequency layer configuration.
Multi-Band Design
Coordinated design across frequency layers: coverage, capacity, and supplementary bands.
02
Data Architecture
A multi-vendor, multi-source ingestion pipeline. Performance, configuration, fault, subscriber-level trace, core-network and field-measurement data all land in one model, alongside the geospatial substrate the RF analysis runs on.
What we handle
Performance (PM)
Ericsson and Nokia 3GPP PM XML (TS 32.435, TS 32.401), benchmarked against 3GPP counter definitions: TS 32.425 for LTE, TS 28.552 for 5G.
Configuration (CM)
Nokia RAML and Ericsson 3GPP CM XML. Full managed object tree with parameter versioning.
Faults (FM)
Ericsson NE3S and Nokia IRP alarm streams, correlated against performance and configuration.
Trace
Ericsson CTR, decoded and geolocated to UE level. Ericsson GPB, Nokia .dat and Huawei .sig decode today, persistence in progress.
Core network (CDR)
Nokia and Cisco SGW and PGW records, for RAN-to-core correlation and subscriber-level outcomes.
Drive test
GNetTrack, MobileInsight and Systemics campaigns. A profile-driven reader takes CSV formats we have not seen before.
Geospatial substrate
WorldPop population density, clutter and terrain surfaces, which the coverage and planning models read.
Streaming (R1 / VES)
PM and CM over VES for networks that publish events rather than drop files.
Synthetic
A generator that plants controlled faults, so a pipeline can be proved before it touches live data.
One model underneath
Many sources, one vocabulary.
Vendor-specific counters, parameters and identifiers are normalised into a single schema on the way in, so an rApp, an agent or a Digital Twin simulation is written once and runs on any network. Adding a vendor is an ingestion job, not a rewrite of everything downstream.
03
PM Insights
Near real-time and historical performance analytics across all cells, sectors, and frequency layers. Configurable KPI formulas, trend analysis, and automated anomaly detection.
KPI Engine
Configurable formulas aligned to operator standards. Coverage, capacity, mobility, accessibility, and retainability domains.
Trend Analysis
Multi-resolution time-series from 15-minute ROPs through hourly, daily and monthly rollups. Detect degradation before outages.
Distribution Analysis
CQI, TA, RSRP, RSRQ distributions per cell. Identify skewed distributions indicating RF or capacity issues.
Counter Explorer
Browse and visualise raw PM counters with filtering by cell, site, band, and time range.
Anomaly Detection
Discover anomalous sites and cells across KPIs via time-series and KPI-encoded clustering.
Cross-Cell Comparison
Compare KPI performance across cells, sites, or clusters to identify outliers.
RAN to Core
End-to-end dependency graph built from CDR, linking cells to SGW, PGW and service. Trace the blast radius of a degradation, faceted by device, TAC, QCI and RAT.
04
CM Insights
Configuration management analytics that surface misconfigurations, inconsistencies, and optimisation opportunities across your entire RAN parameter set.
Configuration Similarity
UMAP-based clustering to identify cells configured differently from their peers without justification.
Parameter Auditing
Compare live CM parameters against golden baselines or vendor best-practice templates.
Change Tracking
Full version history with timestamp, source, and before/after values for audit trails.
Cross-Domain Correlation
Correlate CM parameter states with PM KPI outcomes. Surface the link when configuration change causes degradation.
05
Geospatial Insights & rApps
rApps scan your network continuously, powered by geospatial analysis and geohash-indexed coverage grids. The ones below are where most operators start, not the limit of what an rApp can do.
Start with these
Low Coverage
Clusters of weak coverage grids.
Interference
Geohash clustering overlap analysis.
Overshooting
Downtilt modelling and coverage boundary analysis.
Undershooting
Uptilt modelling for restricted coverage detection.
Silent Cell
Confirmed outages: an active FM alarm with sustained zero PDCP traffic across consecutive ROPs.
Sleeping Cell
Cells underperforming against their own baseline with no FM alarm behind it. Capacity loss nothing else reports.
VoLTE Quality
Voice accessibility, drop-call rate, E-RAB retainability and SRVCC success against sustained thresholds.
Mobility Robustness
Handover and radio-link failures classified into the four 3GPP MRO modes: too early, too late, wrong cell, ping-pong.
PCI Conflict
Same-band cells reusing a PCI with overlapping coverage, causing confusion and handover failure.
PCI Planner
Collision-free PCI plans from handover relations.
Layer Management
Multi-band imbalance: missing band, isolated layer, traffic skew, from HO relations and traffic share.
Layer Mode Loop
Idle and connected-mode thresholds that trap UEs in reselection or mode-boundary oscillation.
CA Imbalance
Coverage versus capacity band traffic imbalance.
Crossed Feeders
Multi-level antenna swap detection from KPI patterns.
Then build your own
The catalogue is a starting point, not the ceiling.
Author your own rApps against the same registry, the same normalised data model and the same Digital Twin that validates everything above. Describe the outcome you want, pick the data scope, tune it on a live preview, and it runs beside the ones we ship. The rApps your network needs are not gated on our roadmap.
06
Agentic Layer
A hierarchical AI agent architecture. Domain-specific sub-agents coordinated through orchestration and management layers. Built on MCP for data access, A2A for inter-agent coordination, and a swappable model provider layer.
Manager Agent
Resolves conflicts between recommendations, enforces operator constraints, and keeps every optimisation in sync across the network.
Orchestrator Agent
Drives the assess-to-fix cycle, routing tasks to domain agents via A2A protocol and synthesising their findings.
Assessor Agent
The quality gate. Scores every proposed change across weighted dimensions: coverage, interference, capacity, mobility, accessibility, retainability and CM configuration.
Fixer Agent
Applies each proposed change in the Digital Twin, measures the KPI impact, and releases only what clears your thresholds, to an engineer or to your SMO.
Domain Sub-Agents
Specialised agents covering coverage and interference, infrastructure, planning, layer management, PM anomaly, PM capacity, and CM insights.
Digital Twin Integration
The one mandatory checkpoint. Every change is simulated against your live network state, and anything that fails to clear your confidence thresholds goes no further.
Architecture
One pipeline, seven stages.
Everything above is one flow. Data arrives raw and vendor-shaped, and leaves as a change that has already been tested. Each stage adds something the next one needs, and nothing further down invents information that was not carried into it.
01
Parse
Many sources, one model
PM, CM, FM, trace, CDR, drive test, geospatial and streaming data land in a single normalised schema. Vendor differences are resolved here, once, instead of in every consumer downstream.
02
KPI enrich
Counters become meaning
Raw counters combine into KPIs by formulas you control, aligned to operator standards across coverage, capacity, mobility, accessibility and retainability.
03
Geo enrich
Numbers get a place
UE-level trace positions and timing advance build geohash-indexed coverage grids, so a KPI stops being one number per cell and becomes a surface you can point at.
04
Insights
The network against itself
Distributions, trends and peer comparison across cells, sites and layers. Configuration similarity surfaces the cells set up unlike their neighbours with nothing to justify it.
05
rApps
Evidence, not opinion
Algorithms with stated inputs, stated thresholds and a stated basis, whether that is RF geometry, a 3GPP procedure or a clustering model. Every finding traces back to what produced it.
06
Agentic reasoning
Findings become a diagnosis
Agents correlate findings across domains, weigh them against your operator constraints, and propose a change with its rationale. Judgement sits on top of deterministic evidence, never instead of it.
07
Digital Twin
Tested before it ships
Hata-COST231 propagation, SINR and KPI projection against your live network state. A proposal arrives carrying its expected outcome, or it does not leave.
Give ARANO a RAN problem
Find it. Understand it. Simulate the fix.
See how multi-vendor RAN evidence becomes a Digital-Twin-validated change, on your data.