It remains unclear how the system reacts to sudden drops in a previously increasing load trend.
Link Load (Figure 6) showed that all schemes (except SFR because of repetitions) have a rapidly decreasing link load trend.
To examine if the predicted values of the response time approach adequately the future values of the load trend.
In our analysis, we evaluated the need to employ load trackers to smoothen the behaviour of the load trend of the network response time.
However, the forecasting methods based on annual system load data are gradually incapable to grasp the load trend in the transitional period mentioned above.
Their disadvantage lies in the introduction of delay in the representation of the load trend, and oscillations when limited set of data are used for resource measurements.
As the focus of this paper is on long-term forecasting, we are proposing a method that would solve the negative peak load trend at the time when DR/DSM programs are introduced and expanded.
In this paper, we are using all historical values from 2001-2014 for the load trend modeling in Step 2. Therefore, Fig. 5 shows a slightly different forecast range compared to that in Fig. 9 in [16].
Andreolini et al. [21] propose an approach for host overload detection in which a host is declared as overloaded when there is a substantial change in the load trend of the host, as a result of applying the CUSUM algorithm.
Under such circumstances, the load trend is changing from the fast growing stage to the saturating and fluctuating stage, which weakens the regularity and makes it difficult to conduct accurate forecasting work.
