Smart Farm Advisory System for Smallholder Dairy Farms in Rural Regions

Thesis Abstract

Smallholder dairy farming in rural regions faces productivity and profitability constraints due to limited access to timely, context-specific advisory information, fluctuating feed costs, disease risks, and poor record-keeping. This study proposes and evaluates a Smart Farm Advisory System (SFAS) designed to deliver real-time, ICT-driven recommendations to smallholder dairy farmers, with the aim of improving milk yield, cow health, and farm profitability. Specific objectives are (1) to design an integrated SFAS architecture incorporating sensor data from milking and housing facilities, mobile-based advisory delivery, and a plug-in decision-support module; (2) to evaluate the system’s impact on milk yield, somatic cell count, feed efficiency, and veterinary expenditure; (3) to assess user acceptance, practicability, and barriers to adoption among smallholder farmers; (4) to examine how farmer demographics, literacy, and access to ICT influence system effectiveness; and (5) to develop a scalable implementation roadmap for rural dairy contexts. A quasi-experimental, mixed-methods research design will be employed. The population comprises 1,200 smallholder dairy farms across three rural districts in a representative tropical region, with a purposive subsample of 300 farms for intervention and 300 as a control group matched on herd size, parity, and baseline productivity. Data collection will combine quantitative instruments—structured farm surveys, sensor-derived metrics (milking interval, cooling uptime, ambient temperature and humidity, feed intake proxies), and farm records—with qualitative tools including semi-structured interviews and focus group discussions with farmers and field technicians. Instrument validity and reliability will be established through content validity indexes, pilot testing (n=30), and test–retest reliability analyses. The SFAS will be implemented for 12 months, delivering tailored recommendations via a smartphone app and SMS, supported by an ontology-based knowledge base anchored in the Theory of Planned Behavior and Diffusion of Innovations theory to explain acceptance and adoption dynamics. Data analysis will utilize regression-based impact evaluation (difference-in-differences with propensity score matching) to estimate changes in milk yield, feed efficiency, and veterinary costs, complemented by time-series analyses on sensor-derived indicators. Thematic analysis will be applied to interview transcripts to elucidate user experience, trust, perceived usefulness, and barriers to adoption. A mixed-methods integrative approach will triangulate quantitative outcomes with qualitative insights to provide a comprehensive assessment of SFAS effectiveness. Expected findings include statistically significant improvements in average daily milk yield (target gain of 8–12%), enhanced feed efficiency (reduced feed costs per liter by 6–10%), lower somatic cell counts indicating better udder health, and a reduction in veterinary expenditures by 10–15% among intervention farms relative to controls. Adoption determinants are anticipated to include perceived usefulness and ease of use, combined with higher ICT literacy and stronger extension support. Sensitivity analyses will test robustness to sensor data quality, network outages, and differing farm sizes. The study will contribute to knowledge by empirically validating an ICT-driven, data-informed advisory model for smallholder dairy systems, bridging gaps in real-time decision support, extension services, and farm-management information systems within rural economies. Theoretical contributions include empirical refinement of the Theory of Planned Behavior and Diffusion of Innovations in ICT-enabled agricultural advisory contexts, and the development of a scalable, ontology-driven knowledge base for livestock management. Practical implications encompass a replicable SFAS architecture, implementation guidelines for rural settings, and policy recommendations on ICT infrastructure investment, farmer training, and coupling with veterinary extension services. The main conclusion posits that a properly designed Smart Farm Advisory System, when aligned with farmers’ cognitive and infrastructural realities, can meaningfully enhance productivity and profitability for smallholder dairy operations in rural regions. Recommendations include scaling the system with modular sensor kits, integrating offline-capable features to mitigate connectivity issues, providing continuous on-farm training, and establishing public–private partnerships to sustain data management and advisory services.