ABSTRACT: RunIt is a proposed mobile application that uses running activity as the basis for a location-based territory game. Instead of treating a run only as a set of fitness measurements, the application gives participants a chance to claim and defend mapped areas by completing runs within them. Geographical space is organised with H3, a hierarchical system that represents locations as hexagonal cells. The project also considers how false location records could affect fair play. To reduce this risk, the proposed verification process examines GPS coordinates, recording times, reported accuracy, changes in speed, and movement continuity. Where device permissions and hardware allow, accelerometer and gyroscope readings may provide further context. Random Forest is proposed for classifying activity records when labelled examples are available. Isolation Forest may be used to flag records that differ markedly from ordinary activity patterns. These model outputs are intended to work alongside basic validation rules rather than act as unquestionable proof of cheating. The paper outlines the application's design, verification workflow, limitations, and evaluation plan. Implementation and testing with genuine runs and controlled spoofing cases are needed before its detection performance can be established.
KEYWORDS: RunIt, location-based gaming, running application, H3, GPS spoofing, activity verification, Random Forest, Isolation Forest, smartphone sensors.
RUNIT: AN AI AND MACHINE LEARNING-BASED TERRITORY-CAPTURE RUNNING APPLICATION WITH GPS SPOOFING DETECTION AND PREVENTION
MR. ARYAN MAHADIK, MRS. ANJALI DANDEKAR


