Positioning systems for underground tunnel environments

In the last years the world has witnessed a remarkable change in the computing concept by entering the mobile era. Incredibly powerful smartphones have proliferated at stunning pace and tablet computers are capable of running demanding applications and meet new business requirements. Being wireless, localization has become crucial not only to serve individuals but also help companies in industrial and safety processes. In the context of the Radiation Protection group at CERN, automatic localization, besides allowing to find people, would help improving the radiation surveys performed regularly along the accelerator tunnels. The research presented in this thesis attempts to answer questions relatively to the viability of localization in a harsh conditions tunnel: “Is localization in a very long tunnel possible, meeting its restrictions and without incurring prohibitive costs and infrastructure?”, “Can one achieve meter-level accuracy with GSM deployed over leaky-feeder?”, “Is it possible to prototype a localization system without a team of hardware engineers?”. To help answering those questions, in the first place, a comprehensive characterization of the power profile in the LHC tunnel was performed for both GSM and WLAN networks, which were transmitted over leaky-feeder cable. Subsequently, several RSSI fingerprinting methods were explored. During the characterization of the power profile, it was noticeable that GSM suffered low attenuation as it propagated in the leaky feeder, at the same time it exhibited significant changes in a short scale and among measurement sessions. Such findings motivated the research of new variants of KNN better suited for leaky-feeder, as well as fusion techniques taking WLAN network in addition. It was found that, even though KNN variants could bring interesting improvements, up to 27%, much more significant gains were attained when considering signals from the WLAN as they exhibited higher attenuation, enabling for 30 meters accuracy in 91% of the cases. To further improve accuracy to the envisaged levels, time-of-flight techniques in narrowband were investigated. A complementary positioning system based on phase delay and aided by synchronization units is proposed and several tests are implemented using Software Defined Radio. Despite the limitations of SDR in achieving phase stability, a method following a round-trip design was shown to correctly stabilize and precisely detect small displacements.