Computational Intelligence-Based Carbon Neutral Wireless Networks for Edge-Cloud Continuum

Big data analytics has major system performance advantages when edge-cloud synergy is available and large datasets can be transferred to cloud computing resources which are superior in computing performance to those assumed to reside in edge servers. To extend this analysis, we explored the total energy requirements of edge-cloud synergy in comparison with using edge computing resources only; we were motivated by the widespread criticism of expanding electrical energy usage by cloud computing resources in consolidated data centres. Using real-world data for power ratings and energy consumption, our simulations show a variable picture of higher energy costs for edge-cloud synergy. These energy penalties are mostly due to high energy estimates for Wide Area Network (WAN) data transfer but projections up to 2030 indicate major improvements in the efficiency of energy use by edge-cloud synergy. This analysis depends on energy per task (in this case 1 GB units of data), which accurately defines energy consumption and the opportunities offered by economies of scale reductions in servers with multiple parallel tasks. The energy penalties are reduced at times of low edge node utilisation or of edge server congestion by over-use and by task partitioning in multiple cloud servers. Any energy penalties associated with edge-cloud synergy are completely eliminated when very high-complexity tasks are offloaded from edge to cloud computing resources even with current WAN energy demand estimates. Our analysis emphasises the operational divergence of enhanced performance metrics and estimates of energy consumption; this can be beneficially exploited to balance differing demands and to set performance targets that take energy use into consideration as a strategy to reduce both edge and cloud energy usage in the transition to net zero climate change mitigation.