Employing a Machine Learning Approach to Detect Combined Internet of Things Attacks against Two Objective Functions Using a Novel Dataset
Table 4
Data preprocessing steps.
Feature
Remarks
Node
Rounded to 0 decimal places
Received
No change
Dups
Feature removed
Lost
No change
Hops
Rounded to 0 decimal places
RtMetric
Rounded to 0 decimal places
ETX
Rounded to 0 decimal places
Churn
No change
Beacon interval
No change
Reboots
Feature removed
CPU power
Converted from W to fW (1 × 10−15)
LPM power
Converted from W to fW (1 × 10−15)
Listen power
Converted from W to fW (1 × 10−15)
Transmit power
Converted from W to fW (1 × 10−15)
Power
Converted from W to fW (1 × 10−15)
On-time
No change
Listen duty cycle
No change
Transmit duty cycle
No change
Avg interpacket time
No change
Min interpacket time
No change
Max interpacket time
No change
Simulation time
Feature removed
Malicious/benign
All benign and malicious activities were grouped and values converted to numeric 1 = Rank and Version Attack 2 = Benign Activity 3 = Rank and Blackhole Attack 4 = Decreased Path Metric Attack 5 = Rank and Sybil Attack
Objective function
All values were converted to numeric 1 = MRHOF, 2 = OF0
