Cyclic transmission window-based bandwidth allocation scheme for asynchronous time-sensitive industrial applications in TDM-PON

Abstract

The industrial Internet has become an important application scenario for time-division multiplexing passive optical networks (TDM-PONs). Time-sensitive (TS) industrial applications, such as control-to-control communications, require a TDM-PON to have deterministic transmission capability. Traditional bandwidth allocation schemes are unable to fulfill this capability; thus we proposed a time-aware deterministic bandwidth allocation (TA-DetBA) scheme to guarantee deterministic delay and jitter in our previous work. TA-DetBA allocates appropriate transmission windows (TWs) to each TS industrial flow in a supercycle based on the flow’s arrival time, cycle, delay, and jitter requirements. However, “awaring” the arrival time of flow relies on precise time synchronization between the TDM-PON and industrial devices, which makes TA-DetBA unsuitable for those TS industrial applications that are asynchronous to the TDM-PON. In addition, the diversity of flow cycles may make the size of the supercycle and the number of allocated TWs large, thus increasing the complexity of the TW scheduling algorithm. To this end, we propose a cyclic TW-based deterministic bandwidth allocation (CTW-DetBA) scheme for asynchronous industrial applications. First, we design cyclic TWs with appropriate cycle and size for TS industrial flows, which not only improves the schedulability of TWs (i.e., the ratio of the number of successfully scheduled TWs to the total number of TWs) but also reduces the complexity of the algorithm. Then, we construct a TW scheduling model based on the satisfiability model theory to schedule the TWs in the PON upstream frame. Simulation results show that the CTW-DetBA guarantees the deterministic transmission of TS industrial flows that are asynchronous to the TDM-PON system. Compared with the fixed bandwidth allocation scheme, the average schedulability of the CTW-DetBA can be increased by up to 62.03%, and the average resource utilization efficiency can be increased by up to 20.4%.