2.1. Academic Developments
In the last few years, many researchers have studied different aspects of the monitoring and control of home energy. For example, a energy monitoring system using Bluetooth and GSM is introduced in [12
]. Such a system consists of a central server that can be managed via SMS (Short Message Service) and that communicates through Bluetooth with a microcontroller, which is in charge of activating and deactivating electric power. A similar system is presented in [13
], although it makes use of Power Line Communication (PLC). The authors take into account the existence of a solar accumulation system, whose non-critical loads are deactivated when the level of the accumulators is too low. The same authors present in [14
] a similar system that controls remotely an Uninterruptible Power System (UPS). As in [13
], PLC interfaces are embedded into the designs proposed in [15
]. Another power outlet system based on PLC is presented in [17
]: it can control the state of the outlets, and it is able to issue warnings about possible power overloads. In [18
], the researchers replace the local controller with a cloud service in order to share resources and reduce costs, as well as to improve remote access for users.
In addition, the literature contains the work of researchers that have devised different smart plugs. For instance, a socket that uses an Ethernet module and an Arduino-Android platform is introduced in [19
]. The researchers indicate that energy savings of 15% can be obtained thanks to their system. An Android-based application is also proposed by Horvat et al. [20
], who present a Bluetooth Low Energy (BLE) solution for controlling appliances and monitoring power consumption. Another intelligent outlet that includes an Ethernet module is presented in [21
]. Such an outlet is designed to work in a network where most of the energy comes from renewable sources.
Furthermore, other researchers [22
] present a home controller system that makes use of a smart plug to monitor energy consumption efficiently using a serial to RS-485 converter to connect to the network. Subsequently, they provide an analysis of the energy consumption of a pilot house, based on a hypothetical scenario. The results present positive impacts on the energy consumption rate. Different methods are applied on emulated data to obtain gains between 10% and 22% in the energy consumption load.
Regarding the different communication technologies, several monitoring systems using a Wireless Sensor Network (WSN) based on ZigBee are described in the literature [23
]. In the case of [23
], the system proposed allows for remote switching off or on and for consulting its consumption in real time through a Graphical User Interface (GUI). Another similar system that uses ZigBee is [26
], which also introduces the concept of reducing the consumption of devices in stand-by mode and uses various sensors to detect when an appliance is consuming more energy than it should in normal operation. Moreover, the authors of [27
] also use ZigBee and include appliance detection via Radio Frequency IDentification (RFID), adding security features and device-specific monitoring capabilities.
Other authors propose the use of Bluetooth as a enabler of the smart management system. An example is [28
], where a smart office is presented that can control the power state of a user’s PC and the switching off the lights through a location-aware approach based on BLE beacons, smart plugs and a mobile app. The experimental results obtained over a three-month period showed average energy savings of 31.9% for the PCs and 15.3% for the lights.
There are just a few developments in the literature that make use of Wi-Fi technology. For example, Thongkhao et al. [29
] propose a plug that employs a low-cost Wi-Fi controller that contains a microcontroller and a Wi-Fi module. Besides, a bi-stable (latching) relay is employed to reach zero consumption in the relay coil when it is stable. The authors compared measurement accuracy with a reference meter and obtained an error of less than 0.5%.
The optimization of demand-based energy planning systems has been analyzed in the literature, as well. For instance, a communications protocol that considers scheduled and real-time devices is defined in [30
]. In such a protocol, one of the devices assumes the role of the master and is in charge of coordinating the rest. Time is divided into slots, in which the devices negotiate the energy that they will be able to use in the next slot. A similar planning is described in [31
], where devices are classified into switchable and non-switchable, and consumption planning is based on pricing in order to reduce the total cost.
Recent research studies have also been performed on the identification of the different home appliances. Ridi et al. [32
] focused on analyzing classification algorithms, including K-Nearest Neighbor (KNN) and Gaussian Mixture Models (GMM), to recognize electric appliances automatically. The authors propose a system based on low-cost smart plugs that measure current periodically and that produce time series that characterize the consumption of an appliance. Thus, such electric signatures can be used to identify the type of appliance in use. Their best combination of features and classifiers shows a 93.6% accuracy. In 2015, the same authors [33
] adopted machine learning approaches for the identification of appliances through their electric signatures. In addition, the researchers presented a database of 450 signatures that contained different brands and appliance models. Another machine-learning approach that allows for detecting configuration changes of smart plug installations is presented in [34
Residential demand is also a thoroughly-studied topic, since it represents a significant portion of the total system load. Thus, residential Demand Response (DR) programs are important from the system operator’s perspective, and Home Energy Management System (HEMS) is an integral part of a smart grid that can potentially enable DR applications for residential customers. A HEMS is responsible for monitoring and managing the operation of in-home electrical appliances, providing load shifting and shedding according to a specified set of requirements. A possible solution for the management in already existing infrastructures is the use of smart plugs. For instance, an example of a smart socket with voltage modulation is presented in [35
]. Such a device is able to support a fully-decentralized voltage service to control the residential electrical loads. Furthermore, Elma et al. [36
] presented a smart plug designed to provide power reduction without turning the plugged device off, since their voltage is controlled automatically to avoid consumption peaks. The smart plug proposed communicates with an HEMS through a gateway that can measure and collect data about current, voltage or power. The communications between the plug and the HEMS interface are carried out using ZigBee. According to the researchers, their system reduces about 18% home peak demand for passive loads through voltage control. A different approach is taken in [37
], where the authors make use of a data acquisition system to control the state of the outlets and to monitor current consumption.
The inclusion of HEMS in existing electrical appliances has also been studied. Tsunoda et al. [38
] proposed a small-sized electrical power sensor that can be easily installed in home outlets. The sensor can measure the power of existing electrical appliances down to 1 W, and it is able to harvest energy from the power line. The power consumption of the sensor is almost 1/100th of that of conventional products.
Several studies about HEMS have been published recently. For instance, an energy consumption schedule for controllable appliances is described in [39
]. Moreover, two different approaches for smart plug scheduling in an HEMS for DR programs with a time-of-use tariff are presented in [40
]. One of the approaches makes use of centralized information received from several smart plugs that communicate with a home automation controller. The second approach is decentralized and runs locally on each smart plug. Another interesting work is presented in [41
], where an intelligent DC power monitoring system is proposed. The system uses open-source software and guarantees that 10%–15% power savings are achieved with proper setting and scheduling. Similar systems based on ZigBee are described in [42
]. Moreover, an integrated solution is proposed in [44
], where the system enables small residential consumers to provide DR services for grid support considering both local energy resources and end-user’s convenience in a real-household environment.
When an HEMS gets smarter and adapts to the surrounding environment, it is called SHEMS (Smart HEMS). An SHEMS can make dynamic adjustments during the operation of home appliances to reduce energy cost, but as a result, it can affect the Quality of Experience (QoE) perceived by the user. This issue is tackled by some SHEMS, like the one described in [45
], where the researchers propose a system that makes use of different appliance usage profiles to harness renewable energy resources and reduce energy cost by scheduling the tasks to be performed during off-peak hours. Another interesting SHEMS is presented in [46
]. Such a system is controlled by an algorithm that optimizes the load scheduling process depending on price and energy consumption. As an example, the researchers model the thermodynamic process of a water heater considering explicitly user comfort as a constraint. Similarly, Jo et al. [47
] focus on Heating Ventilation and Air Conditioning (HVAC) scheduling, incorporating in their model customer convenience whilst minimizing the overall energy cost of electricity and natural gas. Moreover, an interdisciplinary approach to a decoupled DR strategy and a learning-based HEMS is described in [48
]. Finally, it is worth mentioning a human-centric SHEMS that integrates ubiquitous data sensing from the physical and cyber-spaces to infer patterns of power usage dynamically [49
2.2. Commercial Systems
Today, there are numerous commercial smart power outlets on the market. Most of them were designed as plug-in adapters that act as an intermediary between the device and the power source. However, in terms of features, all of them are still far behind the academic systems mentioned earlier.
For instance, Belkin’s WeMo [50
] uses Wi-Fi and allows users to switch on or off devices connected to it from a mobile application. Orvibo’s S20 [51
] is very similar to Belkin’s device and allows for assigning timers to turn off and on power outlets. MyD-Link DSP-W215 from DLink [52
] also makes use of a Wi-Fi network and adds a current sensor that lets users see the real-time consumption of a power outlet from a mobile application. MeterPlug [53
] is aimed at measuring energy consumption, allowing for turning power outlets on and off manually and adding the functionality of stopping the current from flowing when the device is in stand-by mode. SafePlug [54
] can switch on/off the power, monitor current consumption in real time and identify appliances to prevent fires and electrical shocks. Likewise, Edimax’s SP-1101W Smart Plug [55
] uses Wi-Fi, and it also has an embedded power meter. Additionally, it promises savings with insight into how much you spend on each appliance and when it should be switched off (over-budget alerts). Power can be switched off automatically when a user-defined usage limit is reached.
Other devices worth mentioning are MyPlug 2 from Orange [56
], which uses GSM (SMS); PlugWise [57
] and SwannOne’s SWO-SMP1PA Smart Plug [58
], which use ZigBee; MyModlet from Thinkeco [59
] is a Wi-Fi enabled solution; and Ankuoo’s Neo Smart Plug [60
], which makes use of 3G communications. Table 1
compares the most relevant features of the commercial devices previously mentioned. It can be observed that none of them makes use of real-time pricing to reduce energy cost.
Overall, it can be concluded that current commercial devices provide very similar and basic functionality (i.e., they can remotely turn on and off the outlets, monitor energy consumption and set schedulers and timers), but they are a step behind academic developments.
2.3. Analysis of the State-Of-The-Art
After analyzing all of the references mentioned previously, it is clear that most of the work carried out on smart plugs, especially on commercial sockets, has been focused on being able to control the switching on and off of devices remotely. More recently, other features have been added, like consumption monitoring [52
], but only a few actually study the problem of implementing a planning system based on the price of energy to achieve cost savings and/or a reduction of the energy consumption [30
]. However, note that such planning systems analyzed are limited to theoretical simulations: a real-world implementation of specific scheduling strategies has not been found in the literature as is described in this paper. Moreover, to our knowledge, no academic or commercial development has presented so far a hardware prototype that manages in a practical real-world scenario the price values obtained from a public electricity distributor.
Regarding the technologies used, most of the implementations use wired networks or point-to-point communications using wireless technologies. The latest developments use mesh sensor networks through ZigBee. Although ZigBee transceivers have been used extensively for creating WSNs in different fields [61
], they are still relatively expensive, and they require the use of gateways to communicate through IP-based networks. It is worth mentioning that the ZigBee Alliance developed an open alternative called ZigBee IP [64
], which includes IP connectivity through 6LoWPAN (IPv6 over Low-Power Wireless Personal-Area Networks) [65
]. There are several manufacturers that are already selling ZigBee IP-compliant platforms, but it is not as widespread as the original ZigBee. Nonetheless, the price of Wi-Fi modules, like the one used by the solution proposed in this article, enables the creation of an extensible network, adding the possibility of connecting heterogeneous devices such as smartphones or tablets directly to the network.
Finally, no Wi-Fi smart plug that offers self-organizing mesh networking and the auto-configuration characteristics provided by the power outlet system presented in this paper has been found in the literature.