Department of Electrical and Computer Engineering, University of Saskatchewan, 57 Campus Dr., Saskatoon, SK S7N 5A9, Canada.
Cisco Systems Canada, 2123-595 Burrard St., Vancouver, BC V7X 1L7, Canada.
Sensors (Basel). 2019 Sep 19;19(18):4037. doi: 10.3390/s19184037.
This paper presents two methods to optimize LoRa (Low-Power Long-Range) devices so that implementing multiplier-less pulse shaping filters is more economical. Basic chirp waveforms can be generated more efficiently using the method of chirp segmentation so that only a quarter of the samples needs to be stored in the ROM. Quantization can also be applied to the basic chirp samples in order to reduce the number of unique input values to the filter, which in turn reduces the size of the lookup table for multiplier-less filter implementation. Various tests were performed on a simulated LoRa system in order to evaluate the impact of the quantization error on the system performance. By examining the occupied bandwidth, fast Fourier transform used for symbol demodulation, and bit-error rates, it is shown that even performing a high level of quantization does not cause significant performance degradation. Therefore, the memory requirements of LoRa devices can be significantly reduced by using the methods of chirp segmentation and quantization so as to improve the feasibility of implementing multiplier-less filters in LoRa devices.
本文提出了两种优化 LoRa（低功耗长距离）设备的方法，以便更经济地实现无乘法器脉冲成形滤波器。通过使用啁啾分段的方法，可以更有效地生成基本啁啾波形，从而只需在 ROM 中存储四分之一的样本。还可以对基本啁啾样本进行量化，以减少滤波器的唯一输入值的数量，从而减小无乘法器滤波器实现的查找表的大小。在模拟的 LoRa 系统上进行了各种测试，以评估量化误差对系统性能的影响。通过检查占用带宽、用于符号解调的快速傅里叶变换和误比特率，可以看出，即使进行高水平的量化也不会导致性能显著下降。因此，通过使用啁啾分段和量化的方法，可以显著降低 LoRa 设备的内存要求，从而提高在 LoRa 设备中实现无乘法器滤波器的可行性。
