Native Silicon Engineering

Hardware Motion Engine & Sensor Architecture

How GeniSteps achieves pinpoint step tracking accuracy while consuming less than 1% daily battery, running continuously even after app swipe-aways and device reboots.

Battery Impact
< 1%
DSP Bandpass
0.8 – 3.5 Hz
Hardware FIFO
60s Batch
Stride Pacer
~320 ms

1. Android Native Engine (StepTrackerService)

Unlike generic fitness apps that rely on fragile JavaScript timers that are quickly put to sleep by the Android OS, GeniSteps runs a dedicated foreground health service configured with:

  • FOREGROUND_SERVICE_TYPE_HEALTH: Declares official Android 14+ system compliance for health telemetry, ensuring the system scheduler does not throttle continuous tracking.
  • Task Dismissal Resilience: Implements android:stopWithTask="false"alongside onTaskRemoved() fallbacks using AlarmManager. Even if the user swipes GeniSteps out of their recent apps screen, step tracking never drops.
  • Reboot Persistence: A native BootReceiver registers with system startup broadcasts to resume stride counting the instant the smartphone powers back on.
  • Hardware Step Sensor (TYPE_STEP_COUNTER): Taps directly into the dedicated low-power sensor hub silicon with 60-second hardware FIFO batching. The phone's CPU remains asleep while the hardware sensor accumulates strides.

2. iOS Native Engine (CMPedometer)

On Apple devices, GeniSteps interfaces directly with Apple's M-series and A-series motion coprocessors using the native CMPedometer framework:

  • Zero CPU Overhead: Apple coprocessors accumulate and analyze step strides at the silicon level, meaning GeniSteps consumes zero active CPU during background walks.
  • Silicon-Derived Cadence & Pace: Extracts real-time Steps Per Minute (SPM), meters traveled, and exact pace (seconds per kilometer) directly from Apple motion coprocessors.

3. Smart Accelerometer DSP (Fallback & Zero-Latency Layer)

Hardware sensor hubs are power-efficient, but raw batched data dumps can create artificial 4-to-5 step jumps. GeniSteps solves this with a custom digital signal processing (DSP) layer:

// 4th-Order Digital Bandpass Filter

Filters incoming raw triaxial acceleration between 0.8 Hz and 3.5 Hz, the exact biological frequency envelope of human bipedal walking.

Rejects high-frequency engine vibrations, bumps, and accidental pocket jostling with dynamic peak detection and jerk rejection.

4. Fused Motion Arbiter & SmoothStepPacer

When hardware sensor batches dump 4 or 5 steps at once, the GeniSteps SmoothStepPacerdispenses them 1-by-1 at natural biological pacing (~320ms per stride). When reopening the app after hours away, a smooth cubic ease-out animation catches up the counter smoothly from the previous display count to current ground truth.

5. Transit & Driving Mode

Riding in a bus, car, subway, or train? Simply tap the 🚗 Transit Mode toggle in the home header. GeniSteps immediately suspends step accumulation to eliminate vehicle vibration false steps, displaying an amber status badge until you return to your walk.