When constructing hypergraphs for resource allocation in LTE, one method is using the SIR ratio and if its under a specific threshold we connect two vertices through an edge. How is that specific threshold selected?
The first thing any radio engineer has to acknowledge that hardly any path or network in an wireless domain follows a propagation model strictly. It is why so many models (Friis', Hata, Okumura etc...) are used to obtain a consistent result in a specific area.
Now, comes the question of acceptable SIR value for a network. For voice grade networks, the international standard was 12dB min. For data networks, including cellular ones, the threshold is 18dB. For cognitive radios, this threshold may have to be set higher to minimize miss- or false-detection.