Designing a green optical network unit using ARMA-based traffic prediction

Passive optical network (PON) is a popular last mile optical access technology as it offers very high bandwidth to the end customer at a reasonable cost. However, it has been observed that PON is often not energy-efficient and much of its energy consumption occurs from the optical network unit (ONU) located at the user's premises. Suitable protocols can be employed to design energy-efficient ONUs which can operate in different energy-efficient modes. In this paper, we propose a design for green ONUs. Since ONUs are accompanied by buffers that store traffic, the predicted buffer fill-up time can be used as a metric to decide the duration and the type of energy-efficient mode. We have employed traffic prediction to decide the buffer fill-up time. Different energy-efficient modes are employed depending upon their respective time-to-wake-up and energy consumption figures, and the traffic condition. The proposed design does not require any modification to the existing protocols of a PON. This allows operation of the green ONUs over the existing infrastructures.