USRP Based Interference Experiment And LBT Based Throughput Study For LoRa Network

The Low Power Wide Area(LPWA)LoRa technology has found fast-growing popularity for achieving long-range connectivity and low-energy consumption in massive IoT applications.However,the MAC layer of LoRaWAN adopts a protocol similar to non-slotted ALOHA,where the users can send data at any time,resulting in severe collisions and system performance degradation.In addition,the physical layer of the commercial LoRa transceivers is pre-defined,which is not convenient for us to study and implement specific algorithms.In order to tackle this challenge,based on the latest LoRa reverse engineering and the corresponding GnuRadio modules,we successfully rebuilt a point-to-point LoRa communication system,and carried out single-gateway multi-user and multi-gateway single-user experiments,including the behavior of ALOHA protocol that sends data at random time.It is observed from the experiments that when the number of users is large,the system performance will be degraded severely.To improve the multiple access efficiency,we consider the listen-before-talk(LBT)mechanism in the MAC layer.We further investigate how to maximize the overall throughput and achieve fairness of the LoRa network with Poisson randomly distributed users.We derive an accurate and feasible closed-form expression of average throughput of a single user with the spreading factor(SF)s.Based on this formula and the density information of active users,we design an average contention window,and propose an iterative-balance algorithm to allocate spreading factors according to the distance between users and gateway.The Monte Carlo numerical simulations show that our proposed method can maximize the minimum throughput of different spreading factors,namely max-min throughput,and also solve the fairness problem of the network throughput for near and far users.