The standby state can be detected by monitoring the power consumption of the specific device. All these requirements should find agreement in the choice of architecture for the entire system. The latter is the extension of the well-known Internet of Things (IoT) paradigm to the Internet [Guinard et al.
MEASURING ENERGY CONSUMPTION
Systems for Energy Monitoring
Indirect monitoring. As expected, indirect monitoring systems are so named because they do not use electricity sensors to measure the energy consumption of appliances. 2011] contains a proposal on the measurement of energy consumption only for those branches of the energy distribution tree where certain specific devices are connected.
Devices for Energy Sensing
Specifically, for more complex devices with many operating modes, a model of the influence of the magnetic field is used, depending on two previously unknown calibration parameters. It is worth pointing out that hybrid systems are typically characterized by coarse-grained direct monitoring of energy, with a single sensor at the base of the energy distribution tree. However, when the implementation costs are prohibitive, it is possible to reduce the number of devices used and rely on a disaggregation technique, starting from the branches of the energy distribution tree.
Values along the first dimension were assigned to assess both the intrusiveness of installed devices and the discomfort perceived by users during the training phase, while the second dimension is closely related to the position of the assessed solutions within the taxonomy depicted in Figure 6. When it is important to keep installation costs below a given threshold, it will be necessary to trade some of the functionalities in the final BMS for costs. The choice of a given technology directly affects the complexity of the architecture that supports the monitoring system and provides the integration with the rest of the BMS.
Coarse-grained direct energy monitoring systems utilize devices installed at the root of the power distribution network; this represents an extremely simple and inexpensive solution. Monitoring and effectively managing the energy consumption of the sensor infrastructure itself would deserve a separate discussion.
Models of Energy Consumption in Buildings
However, this topic is beyond the scope of this research (the reader is referred to Anastasi et al. [2009] for a detailed review of power management in WSNs). In general, devices consume active power to perform their tasks; however, due to the presence of inductors and/or capacitors in their circuit, reactive power is also consumed. Therefore, a unique device signature is proposed to combine multiple information that can be collected from the power distribution network, such as operating power and power factor.
The use of active power, phase shift, current crest factor and current signal harmonics is proposed in Englert et al. The problem of recognizing appliances in use based on total measurements is also addressed in Ducange et al. This approach differs from previous NALM approaches because it exploits the explicit construction of a model of energy consumption by actuators; however, both in terms of user discomfort and monitoring detail, it suffers from the same limitations as NALM systems.
The energy consumption models discussed so far can also be used to adjust system performance. For this purpose, simulation tools can also be built to model the energy consumption of the entire building [Crawley et al.
ENVIRONMENT AND CONTEXT SENSING
Technologies for Occupancy Detection
Lu and Fu [2009] and Kidd et al. piezoelectric sensors placed under the floor tiles to detect the pressure of the user stepping on. Table II, partially adapted from Lu and Fu [2009], lists some of the available technologies for detecting user presence and activity. Choosing the most suitable one is not immediate, as many factors such as cost, expected performance, intrusiveness and privacy must be considered.
Especially the last two aspects can be of critical importance for the acceptance of the proposed BBS by end users. Privacy concerns are also inherent in dealing with activity tracking, regardless of the sensory technology used. An information fusion process can improve the quality of the information obtained, while maintaining a low intrusiveness.
The selection of the underlying technology requires a preliminary and accurate estimate of the performance to be achieved in terms of costs. On the other hand, the availability of cheaper devices allows the development of a truly ubiquitous sensory system, with broader coverage of the area of interest.
Technologies for Learning User Preferences
Although aiming for a low level of intrusiveness, an optimal choice for the BMS could be the adoption of a wide set of motion and door sensors, so that the data coming from low-cost devices is integrated with that from others beforehand installed devices come, such as noise or power sensors. A detailed review of various information fusion methods for sensor networks is given in Nakamura et al. Such setting makes action monitoring extremely easy, as the necessary information is readily available from the BMS.
2005], where users can only control actuators via an ad hoc panel that later sends all settings to the BMS. Although efficient, the option to have all user-actuator interactions take place through the BMS, thus prohibiting any direct interaction, eliminates the need for traditional control instruments (switches, remote controls), making the whole system hardly attractive to less experienced users , such as the elderly. On the other hand, maintaining traditional modes of interaction reduces the impact on consolidated user habits, at the cost of a higher burden in terms of the technology to be developed and installed, as well as the overall architectural complexity of the BMS.
2008], a feedback is obtained when the actuator state change was not caused by a command from the BMS; a similar approach is also used in Khalili et al. Besides being a source of implicit feedback, user actions also cause changes in the environmental state; it is therefore necessary to handle the possible conflicts between controls generated by the users and by the BMS itself.
INTELLIGENT SUPPORT TECHNIQUES
Occupancy/Activity Detection and Prediction
The first criterion is the overall intrusiveness due to physical devices and user discomfort during the learning phase; the second criterion is the complexity of the adopted energy saving strategy made possible by a suitable set of actuators. The complexity of the software infrastructure makes up for it in terms of a broader set of energy-saving strategies. A straightforward model to derive the current occupancy of a site or the ongoing activities involves the use of statistical correlation of the instantaneous sensor information with the state of interest [Thanayankizil et al.
Furthermore, the various sensory readings are ranked according to a utility index, depending on the relationship between the specific sensor type and the activity to be inferred. To include information about previous states, one of the most common approaches is the probabilistic one, through the use of BNs or more specifically Hidden Markov Models (HMM). 2011]; in the latter case, HMM takes into account both the occupancy level of the room and the estimated number of occupants.
2005], a hidden semi-Markov model is presented, which explicitly takes into account the possible duration of the activities. Furthermore, a low energy saving rate is obtained in the long term due to the user habit of setting the system conservatively to avoid an excessive reduction of their own comfort.
Learning User Preferences
A dynamic learning system allows modification of the user profile as the system receives updated information. When RL is used to learn user preference, usually a negative reward is associated with the last action performed if the user acts. The system uses data mining techniques to detect frequent and periodic patterns in user behavior.
Learning is dynamic and online, and the user can trigger a change in the rules at any time with feedback on the system's performance. 2004] to identify the sequence of actions to be performed on the actuators in order to closely mimic the user's past behavior. In the Neural Network House project [Mozer 1998], user discomfort and energy consumption are considered as two concepts that contribute to the minimization of the same objective function.
2005], expressing it as a linear combination of the user satisfaction function and the cost utility function (which is inversely proportional to the energy saving). Moreover, a predictive model of user presence is embedded in their general model, by considering user preferences only if the probability of the user being in the controlled area is not negligible.
CONCLUSION AND CURRENT CHALLENGES
In the described case study, a solution representing the median of the dominant front is selected. The first issue to consider when promoting the commercial deployment of BMS is an accurate assessment of the return on investment (ROI). Even without considering the costs of software design and development, the actual installation of the necessary hardware (sensors, actuators, communication infrastructure) has non-negligible costs.
The difficulty lies both in the physical deployment of the sensor technology, which could nevertheless require the intervention of technicians, but also in the linking of meta-information to sensors. As can be seen, the majority of intelligent approaches supporting advanced energy saving policies require a learning phase to allow the system to acquire the necessary preparatory knowledge to carry out its own activities. To the best of our knowledge, no work is reported in the current literature addressing the issue of designing a comprehensive system, making full use of intelligent techniques to achieve full autonomy in the control of all aspects of building management .
The authors are grateful to the anonymous reviewers for their insightful comments and constructive suggestions that helped us significantly improve the quality of the manuscript. Enables the use of energy-saving applications on devices in the home environment. IEEE Network.