Automatic Indoor-Outdoor Detection Using Signals of Opportunity

Joshua Sanz, Ali Abedi, Anant Sahai

Abstract: As spectrum sharing becomes increasingly common in bands such as TV whitespace, CBRS, and 6 GHz Wi-Fi, the community is constantly searching for more opportunities where sharing is possible without causing harmful interference. The distinction between indoor and outdoor use has begun to be used, with the underlying idea that indoor use is far less likely to cause harmful interference to an outdoor receiver. Currently, transmitters self-report their location as indoor or outdoor, which is used to determine allowable transmission powers. But how can we reliably check indoor-vs-outdoor automatically? The very crowded nature of radio spectrum that drives the need for spectrum sharing also means that there exist “signals of opportunity” which can be used to sense the radio environment in an approach that we dub “ReciProxy-ADSB.” Aircraft safety broadcasts are a pervasive and useful signal which can provide a radio with directional information about its environment which in turn can produce a more nuanced view of its environment relative to others. We recorded aircraft broadcasts in indoor and outdoor environments then applied model-based and ML-based techniques to classify them. Our preliminary processing shows that with a false positive rate of $30 \%$, we can detect if a station is outdoor with $100 \%$ accuracy. Investigating these false positive cases reveals that in most of these indoor scenarios, there is a nearby window that likely creates a signal propagation environment resembling being outdoors. This important finding should motivate the research community to investigate if the today’s indoor-outdoor classification methodology for interference prevention is effective in practice.