부산대학교 도서관

In the age of digitalization, the impact of Internet of Things (IoT) networks can be seen in many different areas, and there are significant issues with networking and capacity due to the low smartness and cost issues of IoT devices. Congestion and queue management, especially in an IoT network buffer, is one of the most important issues to consider. In this paper, we explained how to develop, implement and test the algorithm built to manage the queue and to avoid potential congestion by adding some intelligence to dumb nodes with a lightweight method called AQM-of- Things (AQMoT) which we have proposed recently. Extensive-form game model is used for defining decision making criteria of both IoT nodes (when to send) and the gateway node (when to drop). We implement a game model according to the queue level by including other network conditions as well. We propose a content-aware priority assignment method for the packets. We develop a realistic test environment and evaluate effectiveness of the AQMoT algorithm. Test results suggest that AQMoT effectively avoids packet drops for high priority data and reduces the burden on the gateway queue caused by redundant data sent by dumb IoT devices.