A Context-Aware IoT Framework for Real-Time Firearm Geolocation and User Authentication: Design, Simulation, and Feasibility Evidence from Kenya
| dc.contributor.author | Mbaluka Solomon Mwaka | |
| dc.date.accessioned | 2026-04-13T08:04:38Z | |
| dc.date.issued | 2026-03-31 | |
| dc.description | Document Type: Master’s Thesis. Institution: Jomo Kenyatta University of Agriculture and Technology (JKUAT), Kenya. Technical Details: Contains 17 pages, 2 figures, and 17 tables. Note on Language: Uses English (US) all through. Related Materials: The simulation code and Random Forest model parameters are available. Status: This is a preprint version of the whole thesis. | |
| dc.description.abstract | Abstract Kenya experiences a critical firearm management gap, evidenced by a 70% surge of illegal firearm recoveries and a reliance on fragmented manual record keeping. This study developed the Proposed Geolocation Mapping Model (PGMM), a context-aware IoT framework designed to transition Kenyan security forces to an automated, real-time oversight. Using a Design Science Research paradigm grounded in Socio-Technical Systems Theory and TAM, the study employed a multi-method approach: thematic analysis of 18 expert consultations, participatory workshops with 24 stakeholders, and machine learning simulations. The PGMM integrates biometric authentication with hybrid GPS/cellular geolocation using dual Random Forest classifiers, Owner Identification (OIC) and Location Identification (LIC), fused via a novel AND-gate decision module. The simulation results identified a high internal consistency, with the OIC achieving 93.0% accuracy (AUC = 0.97) and the LIC achieving 87.5% (AUC = 0.81), both significantly exceeding pre-specified global field benchmarks. A Wilcoxon signed-ranks test confirmed OIC superiority while a Kruskal-Wallis test (p =.048, η2 = 0.41) showed that there existed a significant perception gap, as technology specialists’ ratings implied a high system feasibility than security practitioners’ ratings. These findings show that lightweight IoT data can effectively replace bandwidth-heavy video for firearm management in infrastructure-constrained environments. While the simulation validated the architectural construct, field implementation requires addressing stakeholder perception gaps and doing cellular coverage audits to maintain the LIC’s performance advantage. The PGMM provides a statistically evidenced foundation for transforming firearm management from passive record-keeping to high-assurance, real-time geofencing. Keywords: IoT, biometric authentication, geolocation, Random Forest, firearm management, Design Science Research, decision fusion. | |
| dc.description.sponsorship | This research was self-funded by the author as part of academic requirements at Jomo Kenyatta University of Agriculture and Technology (JKUAT). No external sponsorship was provided. | |
| dc.identifier.uri | https://africarxiv.ubuntunet.net/handle/1/11313 | |
| dc.language.iso | en_US | |
| dc.publisher | AfricArXiv | |
| dc.rights | Attribution 3.0 United States | en |
| dc.rights.uri | http://creativecommons.org/licenses/by/3.0/us/ | |
| dc.subject | IoT | |
| dc.subject | biometric authentication | |
| dc.subject | geolocation | |
| dc.subject | Random Forest | |
| dc.subject | firearm management | |
| dc.subject | Design Science Research | |
| dc.subject | decision fusion. | |
| dc.title | A Context-Aware IoT Framework for Real-Time Firearm Geolocation and User Authentication: Design, Simulation, and Feasibility Evidence from Kenya | |
| dc.title.alternative | Mr. | |
| dc.type | Article |