Table 1 Comparing related works: IDS for attacks on IoT routing

Sinkhole

Le et al. [11]

Use of supernodes, finite machine state, RPL

High positive rate compared to other IDSs for IoT in the literature.

Cervantes et al. [7]

Use of reputation and trust model in INTI, the designed system

Neglect other types of attacks in IoT routing.

Selective forwarding

Mathur et al. [8]

Cryptography

High positive and negative rates, and high energy consumption.

Sinkhole and Selective forwarding

Sheikhan and Bostani [3]

MapReduce, supervised and unsupervised machine learning

Focuses only on static nodes and execution time of approximately 13 min.

Khan and Herrmann [12]

Reputation, trust management

False positive and negative rates above 60%.

Other

Adeilson et al.-2018 [13]

IDS against DoS attacks

Not specific to a type of DoS generating high positive and negative rates

Bhatti et al.-2018 [14]

IDS based on machine learning techniques for detecting network anomalies

Detection not effective for all targeted attacks, and high positive and negative rates.

Sonar et al. [15]

Watchdog on hardware

Limited to a IoT with few nodes and the use of a limiar constraints its effectiveness.

Yang et al. [10]

IDS based on watchdog for detecting false data injection on the network and uses a Bayesian spatio-temporal model

Model yields high energy consumption, and the probabilistic testing applied results in high error rates.