A wireless IoT communication architecture is proposed that takes advantage of RLNC recording properties on natives packets of Core network. The platform aims to cover the shortcomings of the TCP/UDP/IP internet protocol for a traffic streaming intrasession sensitive to the delay in wireless networks since the current systems in native mode do not differentiate between transmission failures due to bottlenecks caused by congestion or by the characteristics of the random medium. The system establishes a mapping of the population of packets to be transmitted over an RLNC coding system based on Finite fields (Galois field), which allows forwarding nodes to eliminate redundant information on the network to decrease communication times and increase throughput. Moreover, thanks to its configuration, the model takes advantage of a Cross-Layer communication, to define a dynamic trade-off between the parameters that define the performance of the system, which translates into a coding density in a reduction of the delay in each of the transmissions. The communication platform is presented under a simulation scheme supported by (KODO and NS3) and is compatible with the Multicast, Unicast, and Broadcast transmission types established under a channel with erasure effect.