Investigation on power efficiency of GPON with heterogeneous traffic

Abstract

In this article, we investigate the performance of Gigabit Passive Optical Network (GPON) with the heterogeneous traffic at Optical Network Unit (ONU) and Optical Line Terminal (OLT) in the presence and absence of watchful sleep mode. OLT located at central office sends the corresponding signals to ONUs at user premises about their next state of operation. In GPON there are three states of ONU namely, active, sleep and doze. The status of transceiver in active state is ON and in sleep state it is OFF, whereas in the doze state the transmitter is OFF and receiver is ON. Three limitations have been observed during the state transitions with active state as an intermediate state between doze and sleep states. First, in doze state, though there is no downstream traffic for transmission, the ONU receiver will be kept ON until the state elapses. Second, during the sleep state, in order to receive the control message from OLT, the ONU will come to the active state despite doze state with its receiver ON which can also meet the requirement. Third, there is no direct transition between doze state and sleep state which leads to an increment in the average power consumption. In order to overcome these limitations, the state transition methodology using watchful sleep mode is introduced, by reducing the transitions with the active state. In this paper, we also suggested another scheme in which ONU transits from the sleep state to doze state by making its receiver only ON, and then from doze state to the active state based on packets availability. We calculated an average power consumption using Discrete Time Markov Chain (DTMC) for GPON with states stationary probabilities. Power saving scheme with watchful sleep mode results in an average reduction in power consumption of 47% compared to the existing method whereas it is 8% for power saving scheme with doze state as intermediate state between active and sleep states. Furthermore, analysis of stationary probability for the active state is also given.