Age of Information in Massive Sensor Networks

In many of the concurrent sensor and IoT networks, information freshness is becoming crucial to the operation of the network. In these networks, receiving the data after long time of being generated (or measured) might be useless. Age of Information (AoI) is a time metric used to investigate the freshness of the received data [1]. AoI is defined as the elapsed time between receiving the last updated version of the data successfully and its generation time at the transmitter [2]. In this proposal, we consider massive IoT networks, where the nodes are randomly distributed, and we evaluate the performance of these networks in the presence of interference from all the nodes to each other. The performance metrics are the achievable rates and the average AoI. We will consider the case of segmented packets, where the transmission of each packet is only successful if all its segments are received successfully. We will consider two scenarios; the scenario of preemptive queues and the scenario of non-preemptive queues. The former scenario ignores the current packet if a new (fresh) packet arrives in the middle of transmission, while the latter scenario ignores the new packet until the current packet is transmitted successfully. It is expected that the performance of the two scenarios will depend on several system parameters such as the success probability (which depends on the interference), the length of the packets, ..etc. Based on that, we can optimize the length of each packet to reduce the average age of information. The tools required to carry on the work above include stochastic processes, stochastic geometry, and optimization. Stochastic processes are needed to mathematically formulate the transmission process of packets with many segments as Markov chains, and calculate the success probabilities and AoI. Stochastic geometry is also needed in order to relate the success probability of each segment to the interference which comes from the nodes that are randomly distributed according to a point process. Finally, optimization is needed in order to optimize the length of the packets to reduce the average AoI