<![CDATA[Learning Analytics (LA) in low-connectivity educational environments remains severely underdeveloped, despite growing evidence that offline-capable infrastructure is critical for equity in the Global South. While a growing body of research explores offline-first and edge architectures for LA, these approaches remain fragmented and lack formally specified, end-to-end pipelines required for deployment in zero-connectivity environments. The dominant LA literature continues to treat reliable cloud connectivity as a design premise rather than a variable, leaving students in rural and marginalized regions without data-driven educational support during offline periods. This systematic literature review (SLR) examines peer-reviewed evidence on the integration of LA with offline-first and edge computing architectures in low-connectivity educational settings. Conducted in accordance with the PRISMA 2020 framework, a structured search across Scopus, IEEE Xplore, and ScienceDirect covering the period 2020 to 2026 yielded 1,539 records; after deduplication, screening, and full-text assessment, 35 studies were included in the final synthesis. Three research questions guided the review: (RQ1) identification of offline-first and edge-based architectural paradigms; (RQ2) evaluation of synchronization and data integrity mechanisms; and (RQ3) assessment of LA algorithms in resource-constrained contexts. Results reveal three validated offline-first paradigms, namely hardware-based local servers, progressive web applications with service worker buffers, and on-device AI inference, alongside consistent evidence that edge architectures reduce latency by up to 71.3% compared with cloud-only configurations. However, no study provided a formally specified protocol for conflict resolution in LA event logs under zero connectivity. This review formally articulates five research gaps (G1-G5) and argues that the formalization of offline-first infrastructure is a foundational prerequisite for both privacy-preserving synchronization and algorithmic advancement in resource-constrained settings.]]>
