Virtual memory systems rely on page replacement policies to maintain performance under resource constraints. Classical algorithms such as FIFO, LRU, Clock or OPT optimize average- or amortized-case behavior, but often fail to ensure fair distribution of page faults across pages. This thesis introduces a Min-Max Paging Framework that focuses on equitable allocation of memory resources by minimizing the worst-case number of page faults incurred by any single page.

The work provides a structured analysis of established paging algorithms through a fairness-oriented lens, characterizes their best-, average- and worst-case behavior, and investigates how adversarial access patterns amplify inequality in page fault distribution. A formalization of min-max fairness in paging is presented, followed by algorithmic contributions and evaluative insights demonstrating how more balanced resource distribution can be achieved. The results highlight conceptual gaps in traditional approaches and outline future research directions toward fair and robust virtual memory management.