This work explores the distributed model predictive control scheme. The Neighbor-Communication, Distributed Constrained-Model Predictive Control (NC-DCMPC) framework is suggested where a centralized problem is divided into dynamically coupled subsystems and the local controllers are allowed to communicate with neighbor agents only. The neighbor structure is defined based on the physical interconnections. Each local controller communicates its predicted effects to its downstream neighbors as well as the predicted costs imposed by the effects from its upstream neighbors. Without any knowledge about neighbor costs or dynamics, converging to the systemwide optimum point is ensured through communication and the closed-loop stability is proved based on this convergence assuming sufficiently a long horizon. To reduce the number of variables due to the long horizon assumption, Laguerre functions are used to parameterize the local control sequences. The result is a smaller size on-line optimization problem for each local controller which allows more communication per sampling time. The proposed algorithm is implemented assuming constraints on the local control actions. An example of networked subsystems is presented to demonstrate the presented DMPC structure.