37 KiB
Estimating the end-to-end energy consumption of IoT devices along with their impact on Cloud and telecommunication infrastructures
Introduction [2 col]
More and more IoT devices: smart building, smart factories, etc. Many sending few data, though taking a lot of bandwidth (Sandvine report) IoT devices, only top of the iceberg: induced consumption on telco net and cloud infra In this paper, we estimate the overall energy consumption of an IoT device environment by combining simulations and real measurements. Sections…
Related Work [1 col]
Smart industry \cite{Wang2016} Smart cities \cite{Ejaz2017}
Use-Case [1 col]
Application Characteristic
The IoT part is composed of an Access Point (AP), connected to several sensors using WIFI. In the system, the IoT part is considered as the part where the system data are created. In fact, the data life cycle start when the sensors records their respective local temperature at a frequency $f$ and the local timestamp. Then, these data are transmitted through the network along with an arbitrary sensor id of 128 bits. Finally, the AP is in charge to transmit the data to the cloud using the network part.
The network part is considered as the medium that link the IoT part to the cloud. It is composed of several network switches and router and it is considered to be a wired network.
Cloud Infrastructure
System Model [2 col]
The system model is divided in two parts. First, the IoT and the Network part are models through simulations. Then, the Cloud part is model using real servers connected to watt-meters. In this way, it is possible to evaluate the end-to-end energy consumption of the system.
IoT Part
In the first place, the IoT part is composed of several sensors connected to an Access Point (AP) which forms a cell. This cell is model using the ns-3 network simulator. Consequently, we setup between 5 and 15 sensors connected to the AP using WIFI 5GHz 802.11n. The node are placed randomly in a rectangle of 400m2 around the AP which corresponds to a typical real use case. All the cell nodes are setup with the default WIFI energy model provided by ns-3. The different energy values used by the energy model are provided on Table \ref{tab:wifi-energy}. These energy were extracted from previous work\cite{halperin_demystifying_nodate,li_end–end_2018} on 802.11n. Besides, we suppose that the energy source of each nodes are unlimited and thus each of them can communicate until the end of all the simulations.
As a scenario, sensors send 192 bits packets to the AP composed of: \textbf{1)} A 128 bits sensors id \textbf{2)} A 32 bits integer representing the temperature \textbf{3)} An integer timestamp representing the temperature sensing time to store them as time series. The data are transmitted immediately at each sensing interval $I$ varied from 1s to 60s. Finally, the AP is in charge of relaying data to the cloud via the network part.
Network Part
The network part represents the a network section starting from the AP to the Cloud excluding the server. It is also model into ns-3. We consider the server to be 9 hops away from the AP with a typical round-trip latency of 100ms from the AP to the server. Each node from the AP to the Cloud is assume to be network switches with static and dynamic network energy consumption. The first 8 hop are edge switches and the last one is consider to be a core switch as mention in \cite{jalali_fog_2016}. ECOFEN \cite{orgerie_ecofen:_2011} is used to model the energy consumption of the network part. ECOFEN is a ns-3 network energy module dedicated to wired network. It is based on an energy-per-bit model including static energy consumption by assuming a linear relation between the amount of data sent to the network interface and its power consumption. The different energy values used to instantiate the ECOFEN energy model for the network part are shown in Table \ref{tab:net-energy} and come from previous work \cite{cornea_studying_2014-1}.
Cloud Part
Finally, to measure the energy consumed by the server, we used real server from the large-scale test-beds Grid5000 (G5K). In fact, G5K has a cluster called Nova composed of several nodes which are connected to watt-meters. In this way, we can benefit from real energy measurements. The server used in the experiment include an Intel Xeon E5-2620 processor with 64 GB of RAM and 600GB of disk space on a Linux based operating system. This server is configured to use KVM as virtualization mechanism. We deploy a classical Linux x86_64 distribution on the Virtual Machine (VM) along with a MySQL database. We used different amount of allocated memory for the VM namely 1024MB/2048MB/4096MB to highlight its effects on the server energy consumption.
The sensors requests are simulated using another server. This server is in charge to send hundred of requests to the VM in order to fill the database. Consequently, it is easy to vary the different requests characteristics namely: \textbf{1)} The number request, to virtually add/remove sensors \textbf{2)} The requests interval.
Evaluation [3 col]
IoT/Network Consumption
In a first place, we start by studying the impact of the sensors position on their energy consumption. To this end, we run several simulations in ns-3 with different sensors position. The results provided by Table \ref{tab:sensorsSendIntervalEffects} show that sensors position have a very low impact on the energy consumption and on the application delay. It has an impact of course, but it is very limited. This due to the fact that in such a scenario with very small number of communications spread over the time, sensors don't have to contend for accessing to the Wifi channel.
Previous work \cite{li_end–end_2018} on similar scenario shows that increasing application accuracy impact strongly the energy consumption in the context of data stream analysis. However, in our case, application accuracy is driven by the sensing interval and thus, the transmit frequency of the sensors. Therefore, we varied the transmission interval of the sensors from 1s to 60s. Some of these results are proposed on Table \ref{tab:sensorsSendIntervalEffects}. In case of small and sporadic network traffic, these results show that with a reasonable transmission interval the energy consumption of the IoT/Network if almost not affected by the variation of this transmission interval. In fact, transmitted data are not large enough to leverage the energy consumed by the network.
The number of sensors is a dominant factor that leverage the energy consumption of the IoT/Network part. Therefore, we varied the number of sensors in the Wifi cell to analyze its impact. The Figure \ref{fig:sensorsNumber} represents the energy consumed by each simulated part according the the number of sensors. It is clear that the energy consumed by the network is the dominant part. However, since the number of sensors is increasing the energy consumed by the network will become negligible face to the energy consume by the sensors. In fact, deploying new sensors in the cell do not introduce much network load. To this end, sensors energy consumption is dominant.
Cloud Energy Consumption
In this End-To-End energy consumption study, cloud account for a huge part of the overall energy consumption. According a report \cite{shehabi_united_2016-1} on United States data center energy usage, the average Power Usage Effectiveness (PUE) of an hyper-scale data center is 1.2. Thus, in our analysis, all energy measurement on cloud server will account for this PUE.
In a first place, we analyze the impact of the VM allocated memory on the server energy consumption. Figure \ref{fig:vmSize} depict the server energy consumption according to the VM allocated memory for 20 sensors sending data every 10s. Note that horizontal red line represent the average energy consumption for the considered sample of energy values. We can see that at each sensing interval, server face to peaks of energy consumption. However, VM allocated memory do not influence energy consumption. In fact, simple database requests do not need any particular huge memory access and processing time. Thus, remaining experiments are based on VM with 1024MB of allocated memory.
Next, we study the effects of increasing the number of sensors on the server energy consumption. Figure \ref{fig:sensorsNumber-server} present the results of the average server energy consumption when varying the number of sensors from 20 to 500 while Figure \ref{fig:sensorsNumber-WPS} present the average server energy cost per sensors according to the number of sensors. These results show a clear linear relation between the number of sensors and the server energy consumption. Moreover, we can see that the more sensors we have per server, the more energy we can save. In fact, since the idle server energy consumption is high, it is more energy efficient to maximize the number of sensors per server. As shown on Figure \ref{fig:sensorsNumber-WPS}, a significant amount of energy can be save when passing from 20 to 300 sensors per server.
A last parameter can leverage server energy consumption namely sensors send interval. In addition to increasing the application accuracy, sensors send interval increase network traffic and database accesses. Figure \ref{fig:sensorsFrequency} present the impact on the server energy consumption of changing the send interval of 50 sensors to 1s, 10s and 30s. We can see that, the lower sensors send interval is, the more server energy consumption peaks occurs. Therefore, it leads to an increase of the server energy consumption.
End-To-End Consumption
To have an overview of the energy consume by the system, it is important to consider the end-to-end energy consumption. The Figure \ref{fig:end-to-end} represents the end-to-end system energy consumption while varying the number of sensors. It is important to see that, for small-scale systems, the server energy consumption is dominant face to the energy consumed by the sensors. However, since we are using a single server, large-scale sensors deployment lead to an increasing consumption of energy in the IoT part. On the other side, network energy consumption is stable regarding to the number of sensors since the system use case do not required large data transfer. Thus, it is important to remember that, to save energy, we should maximize the number of sensors handle by each cloud server while keeping reasonable sensors request intervals.
Discussion [1 col]
Conclusion [1 col]
References [1 col]
\bibliographystyle{IEEEtran} \bibliography{references}