• random access procedure
• nb-iot

As legacy LTE, Narrowband IoT (NB-IoT) provides a contention-based random access procedure to let UEs access the network and obtain uplink resources. System information broadcast by eNB includes the values of the parameters that are needed by a UE to perform the procedure. A change in these parameters may require dozens of seconds for the eNB to inform all UEs in the cell, thus changes should be reduced as much as possible. In order to understand how to set up a stable set of parameters, in this work we analyze the performance of NB-IoT random access procedure in a single cell, in absence of data traffic, assessing the impact of the most important parameters playing a role in the procedure. Past work on this specific topic includes analytical models (based on Markov chains) neglecting some aspects of the system that can be relevant in many scenarios, or high-level abstractions of the procedure, absorbed in a more complex context in which data traffic plays a role in throughput estimation. For this reasons, the analysis is performed through simulations, which also are used, in some cases, to confirm analytical concepts. For this purpose, a simple discrete-event Java simulator has been built from scratch, whose level of abstraction is specifically conceived for random access procedure (not for data traffic). Results show the relationship between arrival rate of UEs and NPDCCH Common Search Space (CSS) period in determining the stability conditions of the system and the role of the parameters (NPRACH and CSS periods, UL repetitions, Response Window and Contention Resolution, maximum number of attempts, backoff) in determining the effectiveness of the procedure, paying attention to UL and DL physical channel occupancy it involves.