Make the evidence repeatable.
Close the remaining technical weaknesses before increasing operational complexity.
Alpine Rescue System is an offline-capable mountain safety platform combining GNSS positioning, LoRa telemetry, emergency signalling and rescue coordination for environments where conventional connectivity cannot be relied upon.
Alpine Rescue System was not conceived as a theoretical classroom exercise. The engineering problem emerged from direct experience of a long mountain journey where distance, time, terrain and limited communications made situational awareness increasingly important.
The resulting question was deliberately narrow: could a low-cost GPS and LoRa prototype provide rescuers with additional structured information — trip context, last-known position and an explicit emergency state — when continuous mobile coverage could not be assumed?
Project-origin journey
Continuous effort
Molodezhny Peak
Registration → telemetry → dashboard
ARS does not attempt to replace certified rescue equipment. It investigates whether a GPS/LoRa MVP can provide additional situational information and support earlier, better-informed decisions.
Mountain groups can move into terrain where cellular coverage becomes unreliable or unavailable. When a group becomes overdue, injured or lost, rescuers may have little recent information about its position, movement history or operational status.
Valleys, terrain obstacles and remote routes can interrupt conventional mobile coverage.
Without recent telemetry, rescuers may lack a reliable final latitude, longitude or altitude.
A missed return time provides limited context when there is no active status or emergency signal.
Alpine Rescue System separates trip registration from radio telemetry. The tourist creates journey context before departure, while the beacon independently transmits live field data. Both streams are joined later in the backend.
Creates identity and route context before departure.
Opens the registration page associated with the selected beacon.
Route, leader, group size, expected return and emergency contact.
The backend validates and stores the new active journey.
Stores deviceId, tripId, route intention and return-time context.
Carries live device evidence independently of the QR workflow.
GPS position, altitude, satellites, battery, status and SOS.
Compact telemetry transmitted without cellular connectivity.
Receives the radio packet and measures RSSI and SNR.
Forwards structured telemetry to the local backend.
One operator view combines who is on the route, where the beacon was last seen, packet freshness, battery state, overdue status and emergency alerts.
OPEN DASHBOARD ↗QR registration does not pass through the beacon or LoRa receiver. It writes trip context directly to the backend. Radio telemetry follows a separate path and is joined later using device and trip identity.
The field prototype is built around the LilyGO T-Beam AXP2101 V1.2 platform. GNSS, LoRa communication, local status, battery support and the primary SOS input are integrated into one serviceable package rather than distributed across multiple external modules.
u-blox positioning provides latitude, longitude, altitude, satellite count and fix information.
Point-to-point telemetry is transmitted to a second matched T-Beam receiver without a cellular link.
Provides local GPS, battery, state and SOS information without requiring a phone or internet connection.
The onboard button uses a deliberate approximately five-second hold to reduce accidental activation.
The MVP deliberately uses integrated, inspectable and replaceable hardware until field evidence justifies custom electronics or a production enclosure.
The enclosure was not treated as decoration. Each iteration responded to fit, access, antenna loading, battery depth, display visibility or durability risk.
The first housing established the overall handheld form, OLED position, antenna clearance and basic internal packaging.
Battery depth, side-wall geometry, OLED opening and antenna orientation were refined around the real T-Beam assembly.
The final MVP revision prioritised readable status, onboard-button access, reduced enclosure penetrations and practical carry geometry.
Bench testing showed that the external LED duplicated information already available on the OLED and dashboard while adding solder joints, separate wiring and another enclosure vulnerability.
The prototype was tested progressively across six field stations with changing altitude, terrain, obstruction and radio geometry.
Initial system verification before moving into progressively more difficult terrain.
Radio behaviour was observed with increased vegetation and partial obstruction.
A transition point between lower terrain and the higher mountain route.
Mynzhylky became an important reference point for higher-altitude and longer-separation testing.
Mountain geometry introduced a more difficult radio path and non-line-of-sight conditions.
The highest and longest recorded test station in the validation programme.
Before departure, the device QR links one physical beacon to one active trip record. This creates the identity, route and expected-return context that later gives incoming telemetry operational meaning.
The QR identifies the selected beacon and opens the registration page for that device.
The group leader enters route, contacts, group size and expected return time.
The backend creates the active trip record and stores the journey context.
Future telemetry can now be interpreted as data belonging to a specific active journey.
QR registration writes directly to the backend. It does not pass through the beacon or LoRa receiver. The beacon carries device identity; the backend stores journey context.
The dashboard combines active-trip information, incoming telemetry, position history and alert state into one operator-facing interface.
The interface is designed around one question: what does the rescue operator need to understand immediately?
Device ID, group leader, route and expected return context identify the journey being monitored.
The latest GNSS coordinates place the beacon on the local map.
Historical telemetry points preserve recent movement rather than showing only one marker.
Reported battery percentage gives the operator additional device-health context.
Radio diagnostics are available for engineering and field-link assessment.
The interface should explain why attention is required rather than displaying colour alone.
New telemetry enters the backend.
Match device to active trip.
Update map and breadcrumb history.
Review time, battery and alert state.
Surface trips requiring operator attention.
The backend converts received serial data into structured telemetry records that can be stored, filtered and visualised.
The dashboard therefore prioritises location, journey state, alert reason and telemetry freshness over decorative information.
ARS did not move directly from idea to finished device. The project progressed through subsystem isolation, mechanical iteration, integration and field validation.
Early development focused on reducing simultaneous unknowns: power startup, OLED output, GPS fix acquisition, LoRa matching and firmware upload were verified before the full system was assembled.
The enclosure was revised around real fit, battery depth, OLED visibility, SMA loading, button access and serviceability.
The final MVP combines the T-Beam platform, GNSS, 868 MHz LoRa, OLED interface, onboard SOS, protected 18650 power, QR identity and the serviceable enclosure into one field-testable system.
Required by the final AXP2101 board revision for stable GPS/OLED operation.
Radio settings were isolated first so communication faults were easier to diagnose.
OLED and dashboard already supplied the information while the LED added wiring and enclosure risk.
Battery depth, antenna loading, OLED opening and button access required physical fit revisions.
Field evaluation was used to identify what information is immediately useful to rescue personnel, what remains operationally weak and what should be changed before any controlled pilot.
GPS position and a clear SOS state were identified as highly useful operator information.
Mechanical robustness remains one of the main barriers before a controlled operational pilot.
Yellow status should communicate the exact warning reason instead of requiring interpretation from colour alone.
Mechanical reliability and field durability affect whether the device can remain operational throughout real mountain use.
Warning colour alone is insufficient when an operator needs to know whether the reason is overdue status, stale data or battery.
False emergency events create operational noise and may reduce confidence in the system.
A fixed receiver location would allow a more operationally relevant controlled range trial.
The approximately 15 km figure is a stakeholder-informed future validation target. It is not a demonstrated ARS range. The current field-tested maximum recorded separation remains 4,180 m.
Feedback becomes engineering evidence only when it is translated into a requirement, prioritised, implemented and retested.
Future development is gated by field evidence. The next version should not add features for appearance; it should close the weaknesses already identified during engineering review and mountain validation.
Close the remaining technical weaknesses before increasing operational complexity.
Only after the MVP is stable should the project test a more realistic multi-user rescue workflow.
Deployment requires much more than hardware. Issue, charging, inspection, training and maintenance become part of the system.
Each next phase is conditional on specific technical and operational evidence.
No unresolved numerical conflict in the validation dataset.
Retesting demonstrates repeatable improvement in weak subsystems.
Roles, alerts and escalation procedures are understood.
Power, enclosure, security and maintenance risks are evaluated.
ARS remains an experimental engineering prototype until repeatable testing, operational approval and production safety controls are completed.
Technical evidence, enterprise planning and multilingual executive summaries are organised as direct project resources.
Hardware architecture, firmware, CAD, enclosure iteration, field validation, failures, engineering decisions and results.
Market logic, stakeholder value, economics, deployment model, pricing assumptions and future operational structure.
Select the language in which you would like to read the ARS executive project summary.
All recorded station outcomes were marked PASS.
Highest recorded station reading at ST05.
Satellites recorded across ST00–ST05.
Separate run from 34% indicated charge to 0%.
Packet delivery varied with terrain, obstruction, antenna orientation and local geometry rather than decreasing monotonically with separation.
GPS remained available at every station. The telemetry pipeline operated from beacon through LoRa receiver, USB Serial, backend ingestion and dashboard monitoring.
ST03 recorded the highest station-level packet loss at 16.7%. Shorter and longer test points sometimes achieved 100%, showing that distance alone is not a sufficient model of radio performance.
Future work requires repeated controlled range trials, reconciled raw Serial and telemetry logs, improved attachment protection and a full 100%-to-shutdown battery test if total runtime is to be claimed.
The 4,180 m figure is the maximum recorded separation in this field programme, not a certified maximum range. GPS availability does not establish metre-level positional accuracy. The endurance test does not establish a complete full-pack runtime from 100%.
ARS uses a three-state operating model to separate normal operation, developing concern and explicit emergency conditions. The purpose is to give the operator more context than a simple online/offline indicator.
The group is within expected operating conditions and no emergency state has been detected.
Something requires operator attention, but the system has not entered an explicit emergency condition.
A direct SOS trigger or explicit emergency condition moves the system into the highest-priority operator state.
The onboard button requires a deliberate hold before an SOS is accepted. The delay is intended to reduce accidental activation during normal handling or transport.
Expected return time has passed.
Device power requires operator attention.
No sufficiently recent packet has been received.
Explicit emergency input has been activated.
State classification is an operator-support mechanism within the prototype. It is not a certified emergency classification standard and should not replace established rescue procedures.
The critical operational chain is designed to remain available locally. Positioning, radio telemetry, serial transfer, backend processing, map rendering and storage can all operate without a live internet connection.
Satellite positioning is independent of mobile data.
Beacon-to-receiver telemetry uses direct radio communication.
Receiver data is passed directly into the local laptop.
Trips, telemetry and events are processed on localhost.
Regional map data is stored locally for field use.
Trip and telemetry history remain available on the laptop.
Hosted dashboard, remote stakeholders and online services.
Beacon, receiver, backend, storage and maps continue locally.
The MVP is designed so that internet access can add convenience and remote visibility, while the core field workflow remains functional on a local laptop.
The beacon transmits the data required to understand where the group is, whether the device is healthy and whether an emergency state has been activated.
OPEN RESCUE DASHBOARD ↗Field testing takes place in the Ile Alatau mountain environment around Almaty.
GPS acquisition, LoRa transmission, SOS logic, battery monitoring and OLED interface.
868 MHz LoRa packets with signal strength, SNR and telemetry delivery testing.
Node.js and Express backend with local telemetry, trips, devices and event storage.
Local Leaflet assets and regional PMTiles map data.
Iterative enclosure development around the T-Beam, battery, display and antenna system.
Live device map, breadcrumb history, overdue tracking, battery state and emergency alerts.
Engineering documentation, field evidence, firmware, CAD iterations and technical material are organised here.
Project overview and stakeholder-facing summary.
DOCUMENTATIONFull architecture, design process and engineering decisions.
ENGINEERING RECORDGPS, LoRa, battery, SOS and offline-system evidence.
FIELD EVIDENCE
Alpine Rescue System is an engineering project developed by Alisher Balbayev and Alina Akylova around one practical question: how can additional rescue information remain available when cellular connectivity cannot be assumed?
Engineering development, system architecture, hardware integration and field validation.
Engineering development, software integration, field testing and project validation.