Frequency Hopping Scheduling Algorithm in Green LoRaWAN: Reinforcement Learning Approach

Long Range Wide Area Network(LoRaWAN) is suitable for wide area sensor networks due to its low cost, long range, and low energy consumption. A device can transmit without interference if it chooses a unique channel, spread factor(SF), and transmission power(TP). In dense networks, the devices run out of unique choices, leading to interference and retransmissions. Eventually, the battery levels of devices drop faster. Also, the selection of transmission parameters affects the Time on Air(ToA) of the signal, thus increasing the energy usage of the end device. Higher values of spread factor, transmission power, coding rate, channel frequencies, and lower values of bandwidth increase energy consumption. Moreover, devices cannot maintain the same values of transmission parameters for long due to duty cycle restrictions. We use two techniques to deal with this situation: Frequency hopping and reinforcement learning. We proposed a scheduling algorithm, ’LoRa-DDPGFHSS,’ based on Frequency Hopping Spread Spectrum(FHSS) and Deep Deterministic Policy Gradient(DDPG) reinforcement learning for dense networks. It schedules device transmission parameters by selecting a unique(channel, SF, TP, time slot) to avoid interference and lower energy consumption. We performed simulations of three scheduling algorithms: FREE [1], LoRaDDPG, and LoRa-DDPG-FHSS using the LoRaSim simulation tool. Our simulation analysis proves that LoRa-DDPG-FHSS improves energy efficiency by about 98% and Time on Air is reduced by 97% compared to FREE in a network with around 4000 devices.