The two classes of medium access control schemes, random access and scheduling, differ in major ways, but they also share many common features. Their differences stem primarily from their very different points of departure. Scheduling techniques have their origins in reservation systems that attempt to emulate the performance of a centrally scheduled system such as a multiplexer. Random access techniques, on the other hand, have their origins in the ALOHA scheme that involves transmitting immediately, and subsequently at random times in response to collisions. The scheduling approach provides methodical orderly access to the medium, whereas random access provides a somewhat chaotic, uncoordinated, and unordered access. The scheduling approach has less variability in the delays encountered by packets and therefore has an edge in supporting applications with stringent delay requirements. On the other hand, when bandwidth is plentiful, random access systems can provide very small delays as long as the systems are operated with light loads.
Both random access and scheduling schemes have the common feature that channel bandwidth is used to provide information that controls the access to the channel. In the case of rescheduling systems, the channel bandwidth carries explicit information that allows stations to schedule their transmissions. In the case of random access systems, channel bandwidth is used in collisions to alert stations of the presence of other transmissions and of the need to spread out their transmissions in time. Indeed, the contention process in CSMA-CD amounts to a distributed form of scheduling to determine which station should transmit next.
Any attempt to achieve throughputs approaching 100 percent involves using some form of coordination, either through polling, token-passing, or some form of contention resolution mechanism. All such systems can be very sensitive to the reaction time in the form of propagation delay and network latency. The comparison of the single-packet and multitoken approaches to operating a token ring shows that a judicious choice of algorithm can result in less demanding reaction times. Truly random access schemes such as ALOHA and slotted ALOHA do not attempt coordination and are not sensitive to the reaction time, but they also do not achieve high throughputs.
Was this article helpful?