In large-scale distributed environments, avoiding concurrent access to the same resource by multiple processes becomes a core challenge, commonly termed distributed mutual exclusion (DME). Token-based mechanisms have long been recognized as an effective strategy, wherein a solitary token is handed around among processes as the key that allows access to the critical section. By doing so, they often reduce the messaging overhead compared to alternate methods. This work surveys the significance of mutual exclusion in distributed computing and examines token-based solutions across various network models (including tree-based, ring-based, fully interconnected graphs, mesh structures, and ad hoc networks). We also delve into essential performance measures such as communication costs and strategies for fault tolerance, then branch into specialized variants, such as k-mutual exclusion and self-stabilizing algorithms. Furthermore, a specialized approach that relies on finite projective planes is introduced to highlight how certain protocols can perform efficiently under both best- and worst-case conditions. Lastly, we explore future directions involving machine learning for token predictive routing and blockchain techniques to resist adversarial behavior. This aims to provide a thorough yet accessible overview of token-based DME approaches, together with insights on emerging research trends.