Techniques for Optimal Exploitation of Green Networking Capabilities in Future Internet Devices

In addition to the description of the network topologies and the devices of the considered reference networks, a detailed profile of their current traffic is required to consider the potential of deploying energy–efficient strategies.

Traffic data for representative links in the TELIT and GRNET network has been collected and the basic characteristics of the extracted traffic profiles are presented. The main result of this analysis is that the network links appear to be under–utilized and that the maximum link occupancy is lower than half of the link capacities in most cases. These results were expected because, as already mentioned, network infrastructures and links are generally dimensioned for at least twice the carried traffic volumes.

Fig. 4.3 compares the maximum, minimum, average and standard deviation statistics with respect to the link capacity for various links in the TELIT and GRNET networks during working days. These statistics represent a normalized value of link utilization, so that they can be compared with each other. The maximum traffic load is less than 55% of the link capacity for every considered network link and is higher than 25% only in a small part of the network links. Hence, the link utilisation is never at maximum capacity, which provides the opportunity to implement energy saving policies according to network traffic.

To individuate the link utilization in the considered network links, Fig. 4.4 shows the probability to have a given level of offered load (with respect to the link capacity) for the working day profile and the holiday profile. The highest probability is for the range between 20 to 25% link utilisation for the TELIT network links and for the range between 5 and 10% for the GRNET network links, followed by the one between 5 and 10%. Note that the differences between the two profiles (working day and holiday) are not significant and these considerations are common to both cases. Because the probability that the offered load exceeds 30% of the link capacity is very low, it is likely that significant energy savings can be obtained by setting the speed of the links based on the actual network traffic.

As shown in Fig. 4.5, the evolution of the traffic offered load has the classical night–and–day profiles. The minimum of the traffic is noticed during the first hours of the morning (from 03:00 to 06:00), while rush hours are during the day. A similar trend is also noticed in case of the holiday profile, as shown in Fig. 4.6.