*6.4. Carbon Footprint*

Carbon footprint can be estimated by using the formula in Equation (9) [128]:

$$\text{CO}\_2\text{e}(\text{g}) = E\_x(\text{KWh}) \times I\_N(\text{g}/\text{KWh}) \tag{9}$$

where *CO*2*<sup>e</sup>* is the number of grams of emitted CO2, *Ex* (*x* equal to *VWW* or *IC*) is the consumed energy (in KWh) and *IN* is the carbon intensity (in grams of emitted CO2 per KWh). This latter parameter can be obtained through the data published publicly by many countries or by organizations such as the European Union, but it is easier to obtain it from Electricity Maps [129], an open-source project that collects such data automatically and plots them through a user-friendly interface. Such a website also indicates the energy sources used by each country (an example of such sources for France, Portugal, Spain, California, and the province of Alberta is shown in Figure 11). The data were obtained for 25 July 2021 and, as it can be observed, energy sources differ significantly from one country to another:


**Figure 11.** Energy sources for France, Portugal, Spain, California, and Alberta (25 July 2021).

Figure 12 shows the estimated CO2 emissions for the energy consumption estimated in the previous section. As it can be easily guessed, emissions increase with the number of deployed mist AI-enabled devices; however, such growth changes dramatically from one country to another depending on the energy source: while near-zero emission countries like France are barely impacted by the increase in the number of deployed devices, a province like Alberta emits more than 17 times more CO2 for 1000 deployed devices.

**Figure 12.** Estimated CO2 emissions for different number of deployed devices for different countries.

It is also possible to obtain the monetary cost of running the mist AI-enabled devices (as an example, the average prices for April 2021 for each territory were considered), which is depicted in Figure 13. As it can be seen in the figure, the cost of running the system in Alberta would be cheaper but will result in more CO2 emissions. In contrast, the countries with the largest shares of renewable energy sources (Spain and Portugal) are the ones with the most expensive electricity. Nonetheless, please note that such a link between the use of renewable energies and cost is impacted by other external factors (e.g., taxes, environmental policy, and energy trading).

**Figure 13.** Electricity cost for different number of deployed devices and for different countries.

#### **7. Future Challenges of Edge-AI G-IoT Systems**

Despite the promising foreseen future of Edge-AI G-IoT systems, it is possible to highlight some open challenges that must be faced by future researchers:


Moreover, the rapid proliferation of new products and devices and their native connectivity (at a global level) will force the convergence of not only G-IoT and Edge-AI, but also 5G/6G communication technologies, the latter being a fundamental prerequisite for future deployments. Indeed, future communications services should also provide better dependability and increased flexibility to effectively cope with a continuously changing environment.

