Collision-resilient LoRa Transmission Using Backscatter

Recent years have witnessed a growing trend in the popularization of LPWAN technologies,which provide long-distance and low-power connection for energy-limited sensors in the Internet of Things.As a representative technology among them,Long Range Radio(LoRa)has received extensive attention from researchers.However,the limited power and low cost of hardware make it difficult for LoRa nodes to deploy complex access protocols,leading to the possibility of collision during transmission among densely deployed LoRa nodes,which may influence the demodulation of symbols.To solve this problem,the existing methods use time windows to segment the collided data symbols by the fact that the data symbols of different LoRa nodes have different time offsets when they arrive at the gateway,and uses the ratios of the peak amplitudes of the segmented symbols to eliminate collisions.Due to the energy loss caused by symbol segmentation,symbols with short length after segmentation are easily submerged by noise,which makes it difficult for the existing methods to be applied in very low signalto-noise ratio(SNR)condition,and limitis the transmission distance of LoRa nodes.Regarding the issue above,a collision-resistant LoRa transmission algorithm supported by backscatter is proposed in this thesis.We deploy several backscatter tags around the LoRa gateway to create unique multipath propagation signatures for each collided data symbol.Then,the original data symbol and the backscattered symbols are combined to demodulate the collided symbols in parallel.Since we don’t need to segment the data symbols,our scheme can decode without energy loss of the original symbol and work under extremely low SNR condition.Specifically:Based on the fact that the chirp modulation in LoRa physical layer provides robust frequency domain characteristics,we focus all the energy of a symbol on the same FFT point,and uses backscattered tags to generate additional peaks in the frequency domain.The peaks of the original symbol and backscattered symbols are combined to identify the symbols from different nodes.Secondly,in order to eliminate the fluctuation of the amplitude of FFT peak,we propose a method to accurately extract the peak amplitude to eliminate the influence of time offset and frequency jump on peak amplitude.Finally,to eliminate the influence of harmonics introduced by backscattered tags on other collided symbols,we make the circuit of the backscatter tags,and proposes a quick phase acquisition algorithm with low time complexity to carry out the iterative recovery of symbols.The experimental results show that,compared with the existing scheme,our scheme can reduce the symbol error rate from 65.3% to 5.5% on average and improve throughput by 15× when SNR is-20 d B.