This paper introduces a new foundation for ultra-cooperative communications based on very low signal-to-noise ratio (SNR) channels between mobile sensor nodes grouped in spatially distributed clusters. In this new framework, we define a cluster entity as a sensor network of wireless nodes with, but not limited to, scale-free topology. We define a physical wireless scale-free model for the clusters that extends the abstract scale-free model developed by Barabasi-Albert for network graphs to the wireless channel. Our air interface model is based on orthogonal frequency division multiplexing with Rayleigh flat fading channels and additive white Gaussian noise. For intracluster communications, we have developed the new concept of consensus in a wireless network framework to improve performance. This new approach is ideal for distributed long-range low power sensor networks with ultra-low power signals well below the noise floor. For a cluster size of 100 nodes, we achieve BER of at an SNR of -25 dB in Rayleigh fading. This method enables reliability in both fading channels for intracluster communications and can support ultra-long range or ultra-low power communications links. We establish the fundamental principles involving the relationship between cluster sizes and communication performance, and illustrate 7× gains in range performance using HATA path loss models.