Adaptive lossless entropy compressors for tiny IoT devices

Internet of Things (IoT) devices are typically powered by small batteries with a limited capacity. Thus, saving power as much as possible becomes crucial to extend their lifetime and therefore to allow their use in real application domains. Since radio communication is in general the main cause of power consumption, one of the most used approaches to save energy is to limit the transmission/reception of data, for instance, by means of data compression. However, the IoT devices are also characterized by limited computational resources which impose the development of specifically designed algorithms. To this aim, we propose to endow the lossless compression algorithm (LEC), previously proposed by us in the context of wireless sensor networks, with two simple adaptation schemes relying on the novel concept of appropriately rotating the prefix-free tables. We tested the proposed schemes on several datasets collected in several real sensor network deployments by monitoring four different environmental phenomena, namely, air and surface temperatures, solar radiation and relative humidity. We show that the adaptation schemes can achieve significant compression efficiencies in all the datasets. Further, we compare such results with the ones obtained by LEC and, by means of a non-parametric multiple statistical test, we show that the performance improvements introduced by the adaptation schemes are statistically significant.