Energies 2015,8 10999 In order to estimate electrical consumption information for individual appliances,intrusive and non-intrusive load monitoring techniques have been widely employed in the related literature [34-41]. Intrusive load monitoring techniques require a meter to be installed at each point of interest (i.e.,at a specific appliance,in a specific office,at a specific receptacle and so forth).However,non-intrusive load monitoring (NILM)techniques rely on the existing available data from the building's electrical meter and employ techniques that identify specific signatures in order to associate energy use with the appliances in operation.In this context,NILM is considered a cost-effective tool to monitor appliance-specific energy consumption,and the current prevalence of NILM indicates its success and feasibility [34,41-43]. It is worth mentioning that the effectiveness of NILM in commercial buildings is quite limited due to the number and abundance of similar appliances in use simultaneously(e.g.,personal computers). Though NILM techniques work at an aggregate scale,there is still a need for effective tools to obtain detailed energy information regarding the consumption behaviors of individual occupants [44].Using individual plug-in level meters in order to find the energy consumption of each occupant at his or her workspace has been used to address this challenge [45,46].One criticism of this approach,though,is that this method is not reasonable in practice as it requires a large initial investment on the part of the business,which thereby decreases the likelihood that companies will adopt the approach.For this reason, researchers have begun looking for alternative means of tracking individual energy use.In their foundational work on this topic,Chen and Ahn [13]attempted to link energy-consuming data with occupancy-sensing data in order to track occupant-specific energy use without the need for capital-intensive plug-in meters.They proposed a coupled system that uses occupants'wireless devices'Wi-Fi connection/disconnection events to collect occupancy-sensing data and then correlates energy-load variations with these events to track occupant-specific energy use.This system confirmed that Wi-Fi connection information could be an effective indicator of energy load variations in commercial buildings.Therefore,this research capitalized on the breadth of research available regarding occupant detection in commercial buildings. Detection technologies typically include cameras [47],CO2 sensors [48],cellular phone control-channel traffic sensors [49],humidity sensors [50],infrared (IR)sensors [51],light sensors [52], motion sensors [53],radio frequency identification (RFID)[54],sound sensors [55],switch door sensors [56],telephone sensors [57],temperature sensors [50],ultra-wideband(UWB)[58],wireless sensor networks (WSN)[59],and Wi-Fi infrastructures [60].These detection technologies can be divided to two main groups [61]:(1)precise technologies with incomplete coverage (e.g.,cameras); and (2)imprecise technologies with full coverage (e.g.,Wi-Fi infrastructures).Cost efficiency, resolution,accuracy,non-intrusiveness,and occupants'privacy are criteria that must be evaluated for occupancy-detection techniques.For instance,some researchers point out that since there are usually multiple overlapping Wi-Fi access points in commercial buildings,Wi-Fi-based occupancy sensing could act as a cost-effective option [13]. In addition,the occupant resolution level of occupancy-sensing is significant for distinguishing the energy-load of a single occupant from a large group of people since the process of coupling occupancy with energy-load data aggregates energy-consumption for all persons within a specified location.There are four levels ofoccupant resolution(see Figure 1)[62]:(1)occupancy:a zone has at least one occupant in it;(2)count:the number of occupants in a zone;(3)identity:who they are;and (4)activity:what theyEnergies 2015, 8 10999 In order to estimate electrical consumption information for individual appliances, intrusive and non-intrusive load monitoring techniques have been widely employed in the related literature [34–41]. Intrusive load monitoring techniques require a meter to be installed at each point of interest (i.e., at a specific appliance, in a specific office, at a specific receptacle and so forth). However, non-intrusive load monitoring (NILM) techniques rely on the existing available data from the building’s electrical meter and employ techniques that identify specific signatures in order to associate energy use with the appliances in operation. In this context, NILM is considered a cost-effective tool to monitor appliance-specific energy consumption, and the current prevalence of NILM indicates its success and feasibility [34,41–43]. It is worth mentioning that the effectiveness of NILM in commercial buildings is quite limited due to the number and abundance of similar appliances in use simultaneously (e.g., personal computers). Though NILM techniques work at an aggregate scale, there is still a need for effective tools to obtain detailed energy information regarding the consumption behaviors of individual occupants [44]. Using individual plug-in level meters in order to find the energy consumption of each occupant at his or her workspace has been used to address this challenge [45,46]. One criticism of this approach, though, is that this method is not reasonable in practice as it requires a large initial investment on the part of the business, which thereby decreases the likelihood that companies will adopt the approach. For this reason, researchers have begun looking for alternative means of tracking individual energy use. In their foundational work on this topic, Chen and Ahn [13] attempted to link energy-consuming data with occupancy-sensing data in order to track occupant-specific energy use without the need for capital-intensive plug-in meters. They proposed a coupled system that uses occupants’ wireless devices’ Wi-Fi connection/disconnection events to collect occupancy-sensing data and then correlates energy-load variations with these events to track occupant-specific energy use. This system confirmed that Wi-Fi connection information could be an effective indicator of energy load variations in commercial buildings. Therefore, this research capitalized on the breadth of research available regarding occupant detection in commercial buildings. Detection technologies typically include cameras [47], CO2 sensors [48], cellular phone control-channel traffic sensors [49], humidity sensors [50], infrared (IR) sensors [51], light sensors [52], motion sensors [53], radio frequency identification (RFID) [54], sound sensors [55], switch door sensors [56], telephone sensors [57], temperature sensors [50], ultra-wideband (UWB) [58], wireless sensor networks (WSN) [59], and Wi-Fi infrastructures [60]. These detection technologies can be divided to two main groups [61]: (1) precise technologies with incomplete coverage (e.g., cameras); and (2) imprecise technologies with full coverage (e.g., Wi-Fi infrastructures). Cost efficiency, resolution, accuracy, non-intrusiveness, and occupants’ privacy are criteria that must be evaluated for occupancy-detection techniques. For instance, some researchers point out that since there are usually multiple overlapping Wi-Fi access points in commercial buildings, Wi-Fi-based occupancy sensing could act as a cost-effective option [13]. In addition, the occupant resolution level of occupancy-sensing is significant for distinguishing the energy-load of a single occupant from a large group of people since the process of coupling occupancy with energy-load data aggregates energy-consumption for all persons within a specified location. There are four levels of occupant resolution (see Figure 1) [62]: (1) occupancy: a zone has at least one occupant in it; (2) count: the number of occupants in a zone; (3) identity: who they are; and (4) activity: what they