Table 1 Literature synthesis detailing methodologies, research objectives, key results, and limitations.
Ref | Method | Main objectives | Findings | Limitations |
|---|---|---|---|---|
AI–IoT smart systems | Plant disease detection, field monitoring, and lightweight deployment in resource-constrained areas | Achieved early disease recognition and precision monitoring through AI IoT integration | Scalability in large farms and false positive rates are not fully addressed | |
Smart urban/Controlled agriculture | Sustainable irrigation, vertical farming, and greenhouse integration | Reduced energy use and validated urban frameworks with Industry 4.0 integration | Limited rural adaptability and incomplete cost evaluation | |
Systematic/ Comprehensive reviews | Survey of CSA technologies, drivers, and service synergies | Mapped AI IoT applications and CSA adoption drivers | Mostly conceptual, lacking causal quantification or experimental validation | |
19] | Edge & Sensor fusion (RNN-LSTM, clustering, 3D sensors) | Energy efficiency, mobility, and adaptive decision-making | Improved energy efficiency, dynamic responses, and object recognition | High computational load and resource-intensive sensors |
Human-centric interfaces (Voice, XAI) | Farmer accessibility and trust in AI | Improved interaction and transparency | Dialect limitations and weak model generalizability | |
Cybersecurity (CTI, Secure Imaging) | Threat detection and visual data protection | Established risk profiling and secure data encoding | No real-time mitigation and computational overheads | |
40] | Deep learning models (CNN, RNN, hybrid) | Crop identification, disease detection, and yield forecasting | Increased accuracy, improved forecasts under climate variability | Hardware dependency, dataset bias, and limited climate re-simulation |
Patent network analysis | Tracking CSA technology evolution | Identified clusters and innovation gaps | No validation of real-world impact | |
Cloud ML quality monitoring | Crop health data integration with healthcare | Linked cloud analytics to food quality | Latency issues in real-time scenarios | |
socioeconomic & policy Studies (Gender, poverty, adoption, financing) | Assessing CSA access, adoption drivers, and economic impact | Demonstrated CSA’s role in poverty reduction, food security, and adoption determinants | Geographic or sectoral constraints, limited scalability, and a lack of longitudinal analysis | |
Business process modelling | Formal verification of Agri-processes | Introduced structured CSA process workflows | Simulation-only validation | |
Water & soil studies (Groundwater, nano-amendments) | Aligning CSA with water-smart practices and soil quality | Region-specific water-smart strategies. nano-bonechar showed potential soil benefits | Absence of predictive modelling and early-stage trials | |
Blockchain and federated learning | Privacy-preserving distributed CSA models | Secured model training with enhanced scalability under climate stress | High computational and communication overhead | |
Time-series rule mining | Sequential agricultural trend analysis | Captured evolving Agri-system patterns | Limited interpretability |
