Table 1.

Literature works and limitations.

Related Works	Limitations
[12,13]	These works do not offer the minimum E2E delay. The reason being that both are primarily based on random placements. Moreover, ED is utilized for propagation delay, which represents real topologies.
[14]	In [14], the authors consider Matlab simulations. However, the results of Mininet are not considered. Moreover, the communication overhead and the results of the controller-to-controller latency are not evaluated.
[17]	An enhanced version of k-means. However, multiple network parameters such as queuing delay are not considered when computing the E2E delay.
[18]	The propagation latency is considered by calculating the geographical distance. However, other metrics are not considered, such as queuing latency, path computation, and link utilization.
[19]	Various factors contributing to the E2E delay are not discussed. The evaluations do not reflect a real network infrastructure.
[21]	Heuristic method of controller placement in SDN. However, queuing and path computing latency are not considered. Moreover, results from Mininet environment are not evaluated.
[22]	A location allocation problem, also known as the NP-hard, but it is not evaluated in the context of SDN.
[23]	A controller placement approach for SD-IoT. However, the results are not evaluated in real Internet topologies. Moreover, there is no comparison of the delay between the switches in the network and the controllers.