This paper proposes a Luby Transform (LT)-like network coding (LTL-NC) scheme for wireless Internet-of-Things (IoT) super-dense network with massive number of nodes and packets as an alternative to the existing networks that require retransmission for the lost or erased packets. The capability of the proposed LTL-NC allows the system not to retransmit the erased packets because the destination can still recover the packets, using other packets received later, in random time-slots. Furthermore, LTL-NC is expected to have low computational complexity and flexibly decode random arrival packets. In this paper, we model both simple and complex wireless networks as well as their random packets arrival using LTL-NC expressed by their different degree distributions. Performances of the proposed LTL-NC are evaluated under frequency-flat Rayleigh fading channels and binary erasure channel (BEC) in terms of packet loss rate (PLR), throughput, and extrinsic information transfer (EXIT) chart obtained from computer simulations. For fairness, wireless networks using the proposed LTL-NC are compared to networks applying existing transfer control protocol/internet protocol (TCP/IP) with the same number of packets and same length of time-slots. We evaluate the performances using a series of computer simulations and confirm the effectiveness of the proposed LTL-NC scheme indicated by higher throughput under the same given network traffic.