Modeling of Sensor Packet Arrival in Simulation Environment

摘要

Today's Internet supports a very varied range of traffic based on the myriad types of applications that are part of the network. Within the last decade, a paradigm known as Internet of Things (IoT) has evolved that has added and continues to rapidly add a multitude of Internet aware devices to the network. Almost simultaneously, the concept of Cloud has emerged having virtually boundless compute and storage capacities and at the same time flexible based on user needs. In the last few years, these two emerging technologies are coming together. With Cloud being the global Internet, IoT promises to add an exciting amount of data to the Internet, both in terms of nature of traffic as well as in volume. Analysis of the traffic is necessary to design network and optimally manage the resources to be provisioned at the service provider end. While it is virtually impossible to analyze every kind of traffic generated by IoT devices, we can classify them based on their general characteristics and analyze each of them independently. In this work, we have selected a simulation based sensor traffic to analyze the distribution of inter-packet times of packets arriving at the receiver using the Cooja simulator. Multiple tests were conducted with varying number of senders from which we find that a high percentage of the packets arrive at the receiver very close to one another and the rest of the packets get spaced apart over time, which fits well with the Pareto distribution. We also see that this behaviour gets more pronounced as the number of senders increase.