Greedy Graph Coloring and Hungarian Algorithms for Resource Scheduling in TWDM-PON

The time and wavelength division multiplexing passive optical network (TWDM-PON) allows numerous users to share a single optical fiber and wavelength. In TWDM-PON, the optical network unit (ONU) has different upstream and downstream user traffic to serve using a limited amount of optical resources. Dynamic wavelength and time slot scheduling play an important role in managing user traffic variation. As is well known, delay is one of the most important performance parameters of TWDM-PON. This motivates us to investigate the delay optimization problem and solve it through resource scheduling. To optimize delay performance, we propose a greedy graph coloring algorithm (GGCA) to dynamically assign the timeslots for upstream packet transmissions. Moreover, we also propose a Hungarian algorithm (HA) to solve the wavelength scheduling problem within the scheduled timeslots by the GGCA. Numerical simulations show that the proposed GGCA and HA provide approximately 70- 80% delay reduction compared to a fixed baseline algorithm that schedules the timeslots and the wavelength in a fixed manner.