We used OMNET++ to conduct our simulation studies. Our main issues were: (i) making reliable connections and (ii) evaluating the end-to-end delay in our converged network. We did a comparative study of both loose and tight coupling. Our network setups for loose coupling and tight coupling are in
shown in Fig. 3 and Fig. 4, respectively
In our simulation we utilized a single gateway that separates
the two heterogeneous networks, that is, the WSN and the LTE-A cellular network. For simplicity, we deployed 10 nodes in our WSN. We did this since we were mainly concerned with end-to-end connections. The reliability of the converged network was limited by the low power, lossy channel of the WSN. Utilizing IEEE 802.5.4 and 6LoWPAN are among the best options for this type of network (Low Power Lossy Network) and were therefore utilized in our simulations. We were also cognizant of potential bottlenecks at the dual mode gateway. This however, was an issue of scalability and could easily be addressed by providing multiple gateways. Nonetheless, issues of scalability were not the focus of our research and hence were not addressed in our simulation studies. The delay at the gateway as a result of buffering and protocol conversion processing was negligible with respect to the delay caused by the noisy channel of the WSN. Thus, this was ignored.
Fig. 6 depicts the end-to-end transmission delay time in loose coupling and tight coupling. The graphs imply that by using the proposed tight coupling method the end-to-end delay time can be decreased significantly from 900 milliseconds at the maximum to 500 milliseconds, which would be a significant enhancement for real time networks or systems with low latency restrictions. Also, this would meet the requirements for various applications of real-time M2M networks.