No abstract

This paper investigates the application of non-orthogonal multiple access (NOMA) and device-to-device (D2D) into the scenario of massive Machine Type Communications (mMTC). Specifically, we first propose a new NOMA-and-D2D integrated network, where NOMA is utilized to deal with the cross-tier and co-tier interference at the base station side. To fully exploit the advantages of the network, we formulate a joint channel allocation and power control problem with the objective to maximize the performance of the D2D communications under the constraints of the rate requirements of the cellular users. For solving the formulated problem efficiently, we first adopt the sequential convex approximation method to solve the channel allocation subproblem, and then transform the power control subproblem into a convex optimization problem. To further reduce the computational complexity, we employ the convolutional neural network (CNN) to devise a resource management framework, where the relation between the system states and the control policies is established by multiple neurons. Finally, simulation results indicate that the convex approximation based algorithm outperforms the other algorithms in terms of utility, sum-rate, and user satisfaction, and the CNN based algorithm achieves orders of magnitude speedup in computational time with only slight loss of performance.

To decrease the scheduling length and energy cost, a distributed link scheduling (DLS) protocol is proposed for Wireless Sensor Networks, which is based on graph coloring. Every node is required to construct its two-hop conflict graph, the scheduling order of every link is decided by its priority and interference degree in the conflict graph. The proposed DLS algorithm relaxes the problem of long scheduling length caused by randomize scheduling and frequent state transition in traditional algorithms. Since DLS can assign adjacent slot for every node, the times of node's state transition and the energy cost can be decreased. The efficiency on decreasing the scheduling length and network energy cost of DLS has been analyzed. The simulation results show that the scheduling length of the proposed DLS protocol is less than DS-fPrIM(Distributed Scheduling-fixed Power protocol Interferences Model) and DRAND (Distributed Randomized time slot scheduling) about 1-2 slots. The scheduling energy cost of DLS is the same as DSfPrIM，less than DRAND. DLS has less state transitions than DS-fPrIM and DRAND about 1 time. The results also indicate that the proposed DLS protocol has good performance on energy efficiency.