Synergizing smart building technologies with data analytics

Andrzej Zarzycki

As interconnected smart devices increasingly augment people’s daily lives, the built environment—the human-made surroundings for people’s lives, including buildings, green spaces, and infrastructure—remains relatively isolated from the pace and the intensity ot this technological progress. The built environment is still heavily grounded in the mindset ot industrialization, with a mechanical frame of reference, and without the benefits of modern sciences and information technologies. While smart technologies are increasingly adopted in cities and buildings (Achten, 2015; Decker, 2017; Zarzycki, 2016), there is still no comprehensive reconceptualization of what the built environment could and should be—rethinking of materiality, systems, and intellectual frameworks. Klaus Schwab points that the fourth industrial revolution—a concept introduced in 2012, characterized by the amplification ot the technological progress by fusing the physical, biological, and digital—is a disruption that affects all aspects of our daily lives (Davis, 2016). This fusion is perhaps the best characterization of the future built environment. Schwab argues convincingly that “The question is not am 1 going to be disrupted but when is the disruption coming, what form will it take and how will it affect me and my organisation?” While these words may sound futuristic, they signal incoming changes to the way the built environment is constructed and used.

Some of these changes manifest themselves through embedded and interconnected devices facilitating new responsive buildings with adaptive assemblies and user-aware behaviors. These smart and autonomous environments are enhanced by the Internet of Things (IoT), a network of interconnected computing devices incorporating embedded systems, real-time analytics, and machine learning. Such environments engage users in an interactive dialogue and are critical tor sustainable practices where buildings respond to environmental factors and monitor their own performance (Akkaya et al., 2015; Coen, 1998). Intelligent systems integrate information technologies with distributed sensing and actuating, raise users’ expectations toward building behaviors, and increasingly address resiliency and zero- energy needs (Gorbil, Filippoupolitis & Gelenbe, 2011; Klippel, Freksa & Winter, 2006; Kobes, Helsloot, De Vries & Post, 2010). They also start to function as active co-participants within the built environment, facilitating occupants’ comfort, safety, and well-being (Corna, Fontana, Nacci & Sciuto, 2015; Li, Calis & Becerik-Gerber, 2012; Schwartz, 2013). These emerging directions associated with information technologies and IoT frameworks not only point to new opportunities tor the built environment but also set new research and design agendas for architects and construction professionals (Wurzer et ah, 2011).

These new opportunities are associated with data collection and analysis as well as embedded intelligence that drives smart buildings. In the current state, many of the smart systems, devices, and appliances function as separate units. While they perform intended tasks well, these systems could also be scaled up and synergized to allow for creating new qualitative data sets and greater understanding of how buildings perform, and occupants behave.

Automated sliding doors in outside vestibules are a good example to illustrate this synergy or a “surplus value” associated with collection and analysis of data coming from this particular unit. While intended to facilitate building access, automated doors could also function as a counting sensor for the occupant movements and loads. This would require a small amount of intelligence added to the automated unit, with the ability to collect and process data as well as to distinguish between groups and individual. However, if the doors were integrated with other intelligent systems, generated data could enhance building operations and safety of its occupants. Knowing the number of people in a building or a space at any given time can facilitate a wide range of building functions and operations. This approach is also scalable and transferable to other similar situations. In the case of the auditorium doors, a counting ability could guard against maximum occupancy code violations or simply provide usage data for more efficient facility operations, such as adjustments to air exchanges or cooling loads due to an increased occupancy, before carbon dioxide (CO,) or temperature sensors would register this change.

This chapter discusses the conceptual underpinnings of intelligence in buildings and construction, with a special focus on the opportunities associated with amplifying current smart technologies through increased interconnectivity with data collection and analysis. It points to the need tor a close integration between smart devices, with the data their sensors produce, and artificial intelligence (AI) tools. Such integration could effectively synergize individual players into coherent and unified “conscious” environments: buildings that are aware ot their performance and occupants. Furthermore, a number of case studies demonstrate how individual building components and assemblies can manifest new functional qualities once interconnected into a network of data-sharing agents. They also show how this qualitative and functional leap can be achieved with low-level building sensing and automation technologies while maintaining effective migration paths toward more intelligent buildings.

Smart building and device characteristics

While smart devices and assemblies can function autonomously when performing complex user tracking or performance optimizations, they do not necessarily need to exhibit the level ot intelligence associated with AI algorithms. The term “smart device” is a rather inclusive category ot objects, from those performing basic automation tasks, such as traditional (even mechanical) sensor-based thermostats or automated tafade systems, to sophisticated predictive devices, such as the Nest thermostat utilizing supervised machine learning. There is little known about specific learning approaches in these commercial products. However, a number of researchers developed similar indoor temperature controlling systems utilizing a wide range of machine learning algorithms, such as a Bayesian-learning with a reinforcement-learning (Q-learning) method (Barrett & Linder, 2015) and the artificial neural network (ANN) (Kazarian et ah, 2017).

The discussion and cited literature point to AI and machine learning playing a critical future role in developing high-level smart environments, particularly those utilizing diverse data sources and interfacing with complex human activities.

The IoT and smart building concepts continue to be a broad and semi-defined territory. Several classification approaches are useful when discussing smart device and smart building frameworks, particularly in the context of framing future developments. Independently ot their level of autonomy and “smartness,” smart objects (buildings, building components, and devices) commonly exhibit the following three typologies or design dimensions proposed by Kortuem et al. (2009):

Awareness is the ability to understand—sense, interpret, and react to—events and human activities occurring in the physical world.

Representation refers to a smart object’s application and programming model—in particular, programming abstractions.

Interaction signals the object’s ability to communicate with the user in terms of control, input, output, and feedback.

The awareness ot smart systems is often achieved through a conditional sequence of sensing (cause), logical processing of data (evaluation), and actuation (effect), diagrammed in Figure 15.1. While the Kortuem et al. classification is rather general, Das and Cook (2005) define smart environment with more granular detail and hierarchical structure, consisting of five layers corresponding to an increased smart object’s intelligence. This classification not only accounts for real-time interactions between devices and users but also identifies the autonomy and decision-making and important features:

  • 1 Remote control of devices: remote and automatic control of devices.
  • 2 Device communication: devices’ communication, data sharing, and information retrieval from outside sources over the internet or wireless infrastructure.
  • 3 Sensory information acquisition /dissemination: information sharing by individual sensors and low-level decision-making.
  • 4 Enhanced services by intelligent devices: includes location and context awareness.
  • 5 Predictive and decision-making capabilities: full automation and adaptation that rely on the Al or information acquisition allowing the software to improve its performance.
Smart objects combine sensing, data processing, and actuation abilities

Figure 15.1 Smart objects combine sensing, data processing, and actuation abilities. The connectivity with other devices creates an opportunity for data sharing and interactions with artificial intelligence algorithms.

As with the classification by Kortuem et ah, any of Das and Cook’s categories could be achieved by individual smart devices or collectively by a group of interacting smart units when sharing and processing acquired data. Currently, only a small number of consumer-grade smart technologies demonstrate the predictive and decision-making capabilities postulated in point 5 above. These include the Nest thermostat with supervised learning and the Google Vision A1Y kit utilizing neural networks. However, the direction ot future developments points decidedly to the role of machine learning and AI in shaping smart building environments.

Building automation

The introduction ot building environmental control systems gave an impulse for the development of early automation controls for indoor air quality (IAQ), temperature, and lighting levels. Though sometimes rudimentary, these controls not only provided increased comfort for living but also improved energy efficiency and reduced the environmental impact of buildings (Guillemin & Morel, 2001). However, automated controls were structured around a central, often isolated, control dashboard with limited sensing locations (e.g., thermostat), each delivering a single data point for a wide number of spatial conditions. This shows the need for more localized and fine-tuned approach to building controls that goes beyond the size of a single room into the scale of individual building components.

A broader building automation platform is needed to scale up data acquisition from sensors and to use these data tor predictive scenarios and decision-making. Currently, building automation is implemented as centralized automatic controls for lighting, heating, and cooling, as well as other building systems including fire, security, and occupant safety. These software platforms are called a building automation system (BAS) and a building management system (BMS). The goals of building automation include improved efficient operation ot building systems, such as reduction in operating costs and energy use as well as increased occupant comfort throughout the life cycle of the building. The majority of currently constructed buildings include BAS/BMS, and many older buildings are retrofitted with these systems. These systems include hardware and software frameworks that integrate controls for all or most building systems within a unified interface (dashboard). Such interfaces are ottered by companies such as Siemens, Honeywell, and Cisco that are already involved in the manufacture ot building environmental system equipment. These systems work effectively, capable of delivering significant cost savings (around 20%) as compared with buildings unequipped with BAS/BMS (Siemens, 2020). However, they usually do not include sophisticated autonomy and intelligence tools. In most cases, they follow a set ot predefined explicit rules rather than respond to historical and real-time data for building occupancy and assembly conditions.

While building automation is seen as the framework behind intelligent buildings, it currently is limited to controlling mechanized and electric/electronic devices, such as cooling and heating systems, without a broader integration of sensors and actuators into building components and assemblies. This is partially because BMS/BAS platforms are developed by companies manufacturing building system components and their controls (heating, ventilating, and air conditioning [HVACJ), not by construction companies or building component fabricators. While these platforms do facilitate performance improvements of installed equipment, this does not address overall building operations or user experience.

This points to the need tor a broader transformation ot the building industry and buildings themselves to integrate technologies, possibly open-source, connecting embedded systems in building assemblies with machine learning tools. Windows, doors, floors, ceilings, and wall panels all could function as part of the building sensing interface interacting with users, collecting environmental inputs, and actuating desired spatial configurations. This would open the smart device ecosystem to machine learning.

A useful analogy to this approach is the European smart city initiative—SmartSantander— where the data from more than 12,000 sensors distributed throughout the city can be freely accessed. The general public is encouraged to use this open-source platform to develop the next level of tools and apps (UCityLabAdmin, 2019). Examples of such uses include a study that correlated traffic patterns with respect to air temperature (Jara, Genoud & Bocchi, 2014) and the use of machine learning and deep learning for parking availability predictions (Awan et ah, 2020). While these examples demonstrate the applicability of machine learning techniques in the context of smart cities, a similar approach could be applied to smart buildings.

Internet of things and building automation

Mark Weiser proposes in the Scientific American article “The Computer for the Twenty-First Century” (Weiser, 1991), “When almost every object either contains a computer or can have a tab attached to it, obtaining information will be trivial.” However, this statement does not consider the impediments coming from the overabundance of information and the limited ability to analyze and act on it. Dealing with large data sets requires robust communication and management systems. In the building context, the IoT can facilitate smart and embedded device communication, sensing, actuation, and interactions with the outside environment. Additionally, smart systems incorporate decision-making abilities by analyzing previously gathered data in a predictive or adaptive manner that often employs AI algorithms. BAS/ BMS platforms benefit from collected building data. However, the IoT framework provides opportunities for greater resiliency and interoperability of the entire system, with data feeds to individual subcomponents—devices and assemblies—as compared to current BASs/BMSs.

While current BAS/BMS platforms follow preprogrammed sets of rules, the expectation is that the underlying reasoning (algorithm) for smart systems would evolve over time based on environmental and user feedback. This feedback could not only enhance smart building operations but also be shared with manufacturers, contractors, and architects. Data shared with clients (BAS/BMS) or interested third parties could be further combined with Al techniques for predictive decision-making regarding building occupancy and usage patterns (Zarzycki, 2018). Another direction for the enhanced data sharing involves connecting building automation and IoT with building information modeling (BIM) to improve operational and construction efficiencies (Marble, 2018; Shelden, 2018; Tang et al., 2019).

Enhanced living

Smart technologies not only address building performance and operations (Pena, Meek & Davis, 2017) but also increasingly focus on human interactions by tracking building occupants (Li, Calis & Becerik-Gerber, 2012) and considering their individual preferences. Through their user-centric focus, adaptable and embedded environments promote more inclusive and diverse groups of occupants by reducing physical accessibility barriers and facilitating independent living (Delnevo, Monti, Foschini & Santonastasi, 2018; Domingo, 2012). People with limited mobility or with visual impairments can greatly benefit from autonomous and smart buildings. For example, ambient assisted living (AAL) relies on smart technologies to support independent living tor seniors in situations that would otherwise require traditional assisted living arrangements. These technologies can monitor seniors’ activities to provide early warning signs and response to bodily activities. They can also facilitate interactions within the built environment, extend sensory perceptions, and simplify performance of daily tasks. As an area of early adoption for smart technologies and machine learning, AAL offers opportunities for a significant payback due to a natural fit between the technologies’ capabilities and users’ needs.

Monitoring and detection of daily activities in the context of AAL has been studied by a number of researchers. Chernbumroong et al. (2013) developed methods for daily activity recognition with a high classification rate exceeding 90% for elderly users utilizing a non-stigmatizing and nonintrusive (nonvisual) wrist sensor-based device. The study identified and tracked nine distinct activities, including eating with utensils, brushing teeth, dressing, walking, sleeping, watching TV, and ascending and descending stairs. In another study, Amoretti et al. (2013) adopted machine learning using classifiers based on a Bayesian network approach (Weka library) for sensor data synthesis, activity monitoring, and context reasoning. Their user activity monitoring system enabled machine learning to train for human posture detection, posture classification, and association of these postures with the activities being performed.

The above studies demonstrate how data generated by sensors and IoT devices can benefit from machine learning and Al techniques by facilitating an in-depth understanding of human activities. These examples indicate broader opportunities for using sensing and wireless communication to increase the built environment’s responsiveness to its occupants. These studies also demonstrate strong benefits of Al tools when working with complex tasks and diverse user groups, particularly those who are physically disadvantaged. Smart technologies can facilitate the next level of accommodations toward people with disabilities and address their needs more directly and individually than is currently done through the Americans with Disabilities Act (ADA). The ADA, providing design guidelines for the static environment, relies on the statistical models of an average person and ergonomic needs associated with particular ailments. On the other hand, smart environments—sensing and adaptive— provide an opportunity to tailor spatial and accessibility requirements to individual users.

The future take on the ADA will require not only hardware solutions, both static and adaptive (e.g. ramps, stairs, areas of refuge, and adjustable sinks), but also embedded solutions (intelligence) where buildings are aware of their occupants and can effectively help them to use, navigate, access, and safely exit spaces (Nagy, Villaggi, Stoddart & Benjamin, 2017; Schwartz & Das, 2019; Simondetti & Birch, 2017). To achieve this, smart environments— building, assemblies, and appliances—should not only adapt in real time to user needs but also power Al frameworks to provide a better understanding of user needs and behavior. By doing so, they can also help to refine future buildings standards and regulations, such as the ADA.

Emerging synergies

A transition from the traditional and mechanical paradigm to electronic and digital presents us with new opportunities—synergies emerging from interconnectivity and a wide range of sensory inputs. It also puzzles us, not knowing how to make that step, how one technology can naturally transition into a new one in a natural, evolutionary way.

When replacing traditional tungsten with the LED light bulb, we not only save energy and have greater control of the light rendition but also open ourselves to new synergies. An LED light bulb no longer has the simplicity of the tungsten incandescent bulb and requires

IS.2 The evolution of a light bulb from incandescent to the LED opens new opportunities for embedded devices

Figure IS.2 The evolution of a light bulb from incandescent to the LED opens new opportunities for embedded devices. An introduction of the circuit boards packed with electronic components allows for additional sensors, actuators, and wireless communication.

a number of additional electronic components to operate (Figure 15.2). While this adds additional cost to the LED light bulb, the investment in the electronic board makes it easy to scale it up and add additional sensing and actuation capacities. This has resulted in various functional variations of a light bulb that emerged recently. From light bulbs controlled with embedded motion or light sensors to those with built-in Bluetooth speakers, these variations exemplify the concept that once even a simple device such as a light bulb gets embedded with a little electronics sensing, actuation, and wireless communication, its form and function open to a wide range of opportunities.

This story of the light bulb is not unique. It parallels similar evolutions of phones, and soon cars and buildings. However, this example points to synergistic opportunities associated with embedded technologies—sensor, actuation, and wireless communication. What is missing from these examples is the next step in integration of digital technologies— intelligence. While the embedded components expand the functionality of the original device or appliance and make them ‘smart,’ they still follow mechanical indiscriminate logic. While they provide an increased performance, often with reduced environmental impact, they cannot address individualized human needs. To realize a broader and more tailored impact, these devices need to be networked and to output data that can be further analyzed and synthesized.

A significant amount of research into smart environments, such as studies associated with AAL, relies on custom-developed sensing devices such as wearable electronics. However, another strategy utilizes existing appliances and sensor technologies. A number of studies use CO, sensors for measuring building occupancy (Pedersen, Nielsen & Petersen, 2017; Zuraimi et al., 2017). This is an example of synergies discussed earlier, when data acquired for one reason can be repurposed for another arena. CO, sensors are increasingly used on the smart and high-performance buildings with reduced HVAC loads that rely on natural ventilation, particularly associated with doors and operable windows (Pena, Meek & Davis, 2017). With tight building envelopes and reduced air exchanges, the accumulation of CO, could significantly impact the IAQ and the well-being of occupants. The doubling up on the use of sensors, such as using CO, for indicating IAQ or measuring occupancy loads, or using multiple sensors to capture complex conditions and human activities (Amoretti et al., 2013), brings to the forefront the need for intelligent data analysis—machine learning and other Al tools. Sensors alone usually do not provide enough reliable data, as they are often triggered by a broad range of inputs and conditions. The postsensing filtration, triangulation, and interpretation of sensor data are often required.

Extended applications

The following case study was developed to test the ideas discussed earlier, specifically emerging synergies from interconnected smart light fixtures. Noninvasive motion detection with the use of passive infrared (PIR) and microwave radar sensors is a common approach to lighting controls and energy savings in the built environment. It provides a high level of reliability with low initial cost and an easy integration with the existing fixtures. While the initial application of this technology was in on-demand lighting controls, mostly for safety and security reasons, they quickly became a standard tor lighting controls in spaces with low or infrequent use. Currently, many office and residential spaces use them as the main lighting control approach. These motion sensors can be directly integrated with light fixtures, providing highly localized controls, or can be placed separately to control a group of fixtures. In the majority of current applications, motion-based lighting controls work autonomously and independent of each other, without data collections or sharing. Fixtures do not interface among each other, nor with a building automation system, since their primary function does not require a networked connection. However, this lack of interconnectivity can be seen as a missed opportunity if one considers possible synergies with other systems. If the wireless networking were implemented, the data could be collected and analyzed, which could facilitate qualitatively new applications.

Counting occupancy during emergencies

The idea behind this project was to channel unutilized sensor data from light controls to develop an occupancy load and space utilization framework that could be used not only for day- to-day facility management operations but also tor emergencies. The knowledge of the state of occupancy—intensity and distribution—just before an emergency occurred (e.g., a fire or an earthquake) would allow more direct rescue operations. This sensor-derived knowledge could further facilitate egress by directing evacuees to fire-separated zones. Machine learning could also help predict possible exit paths utilizing sensor data on how individual users entered the building. Life safety research demonstrates that in case of emergencies, most people do not travel to the closest exit, but instead escape via familiar routes or try to retrace their original arrival sequence (Kobes et al., 2010). In residential and workplace building types, occupants may be familiar with floor layouts and have a good understanding of evacuation routes reinforced by periodic fire drills. By contrast, in buildings serving a large number of frequently changing occupants, such as medical facilities, stopping malls, or transportation hubs, users do not have preexisting knowledge of the building and its egress routes. Additionally, not all occupants are able to leave the building on their own during emergencies, since they would have to rely on elevators. The ability to correlate occupants’ current location with their arrival route and to understand whether they were able to safely exit or reach an area of refuge could provide useful information to first responders. From the smart building perspective, the occupants’ lack of the building layout and egress knowledge could be supplemented with the building’s awareness of its occupants and the ability to guide them to a safe exit.

The project prototype used a web-based interface for occupancy visualization (Figure 15.3). Depending on the filter settings, one could see the most recent occupancy state

Building occupancy visualization in support of first responders

Figure 15.3 Building occupancy visualization in support of first responders. Tracking building occupants not only facilitates an understanding of the space utilizations but also can help first responders in directing their efforts during emergencies. This provides opportunities for data analytics and machine learning tools to enhance building services and occupant well-being.

or an occupancy average over a given period of time tor each space. While the technological setup is rather simple, the ability to collect and analyze data provides new opportunities, such as life safety and first responder support, beyond the original intent of energy efficiency and lighting controls.

Window operation tracking

Similarly, to the light controls in the previous case study, security controls on doors and operable windows could be synergized to enable new understanding of building operations. The second project utilized magnetic switches, commonly used for monitoring doors, windows, and cabinetry to study user overrides of indoor air/climate controls. In the traditional HVAC approach, operable windows were seen as unnecessary feature, since their use interferes with centrally controlled air conditioning and ventilation. By contrast, many contemporary high-performance buildings rely on operable windows to provide required air exchanges and air conditioning while reducing the building’s heating and cooling loads (Pena, Meek & Davis, 2017). However, the use of operable windows is not meant to eliminate the building’s mechanical systems but rather to reduce the reliance on them. If not used properly, operable windows can negatively impact overall building performance. This is why window controls with user overrides need to be closely monitored to strike a balance between energy performance, user comfort, and the user’s need to feel that they are in the control of their environment.

The study looked at the frequency with which operable windows are opened and closed by building occupants and the spatial distribution of these activities. It correlated these events with the temperature measurements inside and outside the building to look for patterns that would explain users’ behavior (Figure 15.4). Since the space was semipublic (an architectural

Opening and closure for three windows correlated to outdoor temperature

Figure 15.4 Opening and closure for three windows correlated to outdoor temperature.

Tracking user actions against building environmental controls allows for an understanding of miscalibrations and can server as the postoccupancy evaluation. The study was developed by Evelin Taipe, Andrew Rivera, and Andrzej Zarzycki. The outside weather data was collected from Weather Underground repository (

studio space shared by about 25—35 students), there was a certain inertia in students’ responses to window operations. In most cases, users acted when the response was necessitated by one of the environmental factors. However, it was difficult to understand to what extent occupants used window controls to modify the quality of indoor air and to what extent these were habitual actions unrelated to the building’s indoor environment. Sensors showed windows being opened during the middle of the day and closed from the evening till the following morning with the exception of one event. With the outside temperatures ranging from low 50’s at night to the mid-70’s during the day the window opening pattern would be consistent with maintaining a low temperature gradient between outside temperature and indoor set at around 70°—72° F. One event shows a window being left open for the entire night. Unfortunately, this study did not measure the occupancy levels, so it was difficult to see if there are other correlations. Studies like this one point to synergies between individual smart building monitoring and environmental control systems for real-time decision-making. They can also be used tor qualitative postoccupancy evaluation (l’OE) of buildings, with findings informing future equipment commissioning and providing feedback to architecture, engineering, and construction teams (Council, 2002; Hardin, 2018; Hiromoto, 2015). An important part if this feedback loop involves sensor data set and machine learning models. For the building energy consumption, ANNs and support vector machines (SVMs) are the most commonly used machine learning methods (Zhao & Magoules, 2012). For indoor thermal comfort predictions, a number of research projects use autoregressive models with exogenous inputs

(ARX), ARMAX and non-linear artificial neural network (ANN) models (Patil, Tantau & Salokhe, 2008; Thomas & Soleimani-Mohseni, 2007) and a back propagation neural network (BPNN) based on principal component analysis (PCA) (He & Ma, 2010).

Postoccupancy evaluation in smart buildings

When commissioning a building, initial equipment and system configurations—such as the calibration of user overrides in operable windows—are based on past, often uncodi- fied, experiences. While these assumptions were informed by past projects, a new location, program type, or occupants often result in the need for a slight variation from the default configuration. An introduction of intelligence and machine learning allows for adaptive adjustment of the smart building’s operational parameters to meet the evolving needs of its occupants—as habits change, building operations change. These adjustments could complement, or perhaps replace, traditional POE that is meant to provide validation tor ordinal design and reinforce research feedback (knowledge loops) advocated by Thomas Fisher (2017) as part of architectural practice. Smart buildings, assemblies, and systems interfaced with AI and machine learning cannot only increase the fit between buildings and their occupants but also feedback this information to clients, architects, and contractors to inform their decision-making. This is evident in the WeWork’s pioneering use of machine learning to evaluate its real estate holdings. The company deploys artificial neural networks to evaluate meeting room usage (Bailey, Phelan, Cosgrove, & Davis, 2018; Phelan, Davis & Anderson, 2017) and SVMs (scikit-learn toolkit) to predict marketability of offices (Fisher, 2017).


Smart buildings and autonomous spaces are at the forefront of the current architectural and design discourse (Park, 2017; Zarzycki & Decker, 2019). Future adaptive architecture will integrate AI with information technologies and distributed sensing to address emerging environmental and human needs. AI will also redefine the role autonomous spaces play as active co-participants in the built environment.

At the same time, smart environments are quickly becoming our social, cultural, and well-being sensors. Like mobile phones, smart buildings not only increase occupants’ comfort and work efficiency but also function as monitoring devices. They collect location and activity data to facilitate better services. While listening to occupants’ needs, they learn their personality, preferences, and daily routines. These data also tell the story of its users. The more individualized experience we expect, the deeper probe of sensors and data analysis we have to accept. While this may feel uncomfortable, how this data is used, sanitized, and shared to maintain integrity and privacy will speak to the success of these technologies.


Achten, H. (2015). Closing the Loop for Interactive Architecture-Internet of Things, Cloud Computing, and Wearables. In Martens, B., Wurzer, G., Grasl T., Lorenz, W. E. and Schaftranek, R. (eds.), Real Time — Proceedings of the 33rd eCAADe Conference (Vol. 2, pp. 623—632), Vienna, Austria.

Akkaya, K., Guvenc, I., Aygun, R., Pala, N., & Kadri, A. (2015, March). IoT-based occupancy monitoring techniques for energy-efficient smart buildings. In 2015 IEEE Wireless Communications and Networking Conference Workshops (WCNCW) (pp. 58—63). IEEE, WCNCW.2015.7122529

Amoretti, M., Copelli, S., Wientapper, F., Furfari, F., Lenzi, S., & Chessa, S. (2013). Sensor data fusion for activity monitoring in the PERSONA ambient assisted living project. Journal of Ambient Intelligence and Humanized Computing, 4, 67—84. Awan, F. M., Saleem, Y., Minerva, R., & Crespi, N. (2020). A comparative analysis of machine/deep learning models for parking space availability prediction. Sensors, 20(1), 322.

Bailey, C., Phelan, N., Cosgrove, A., & Davis, D. (2018). This room is too dark and the shape is too long: quantifying architectural design to predict successful spaces. In De Rycke, K., Gengnagel, C., Baverel, O., Burry, Mueller, C., Nguyen, M. M., Rahm, P., and Thomsen, M. R. (eds.), Humanizing Digital Reality (pp. 337—348). Springer, Singapore.

Barrett, E., & Linder, S. (2015) Autonomous HVAC control, a reinforcement learning approach. In Bifet A. et al. (eds.), Machine Learning and Knowledge Discovery in Databases. ECML PKDD 2015. Lecture Notes in Computer Science, vol. 9286 (pp. 3—19). Springer, Cham. https://doi. org/10.1007/978-3-319-23461-8_1

Chernbumroong, S., Cang, S, Atkins, A., & Yu, H. (2013, April). Elderly activities recognition and classification for applications in assisted living. Expert System with Applications, 40(5), 1662—1674, Apr. 2013. [Online], http://dx.doi.Org/10.1016/j.eswa.2012.09.004 Coen, M. H. (1998). Design principles for intelligent environments. In C.R.A.J. Mostow, (ed.), Proceedings of the Fifteenth National Conference on Artificial Intelligence (AAAI’98) (pp. 547—554), AAAI Press / MIT Press, Madison, Wisconsin.

Corna, A., Fontana, L., Nacci, A. A., & Sciuto, D. (2015, March). Occupancy detection via iBeacon on Android devices for smart building management. In 2015 Design, Automation & Test in Europe Conference & Exhibition (DATE) (pp. 629-632), IEEE. Council, F. F., & National Research Council. (2002). Learning from Our Buildings: A State-of-the-Practice Summary of Post-Occupancy Evaluation (Vol. 145). National Academies Press, Washington, DC.

Das, S. K., & Cook, D. J. (2005, December). Designing smart environments: A paradigm based on learning and prediction. In International Conference on Pattern Recognition and Machine Intelligence (pp. 80—90). Springer, Berlin, Heidelberg.

Davis, N. (2016) What is the fourth industrial revolution. World Economic Forum, https://www.wefo- (accessed July 30, 2020) Decker, M. (2017). Soft human computer interfaces. In Fioravanti, A., Cursi, S., Elahmar, S., Gar- garo, S., Loffreda, G., Novembri, G., Trento, A. (eds.), ShoCK! Sharing Computational Knowledge! — Proceedings of the 35th eCAADe Conference (Vol. 2, pp. 739—744), Rome, Italy.

Delnevo, G., Monti, L., Foschini, F., & Santonastasi, L. (2018, January). On enhancing accessible smart buildings using IoT. In 2018 15th IEEE Annual Consumer Communications & Networking Conference (CCNC) (pp. 1-6). IEEE. Domingo, M. C. (2012). An overview of the Internet of Things for people with disabilities. Journal of Network and Computer Applications, 35(2), 584—596. Fisher, T. (2017). Research and architecture’s knowledge loop. Technology Architecture+ Design, 1(2), 131-134. Gorbil, G., Filippoupohtis, A., & Gelenbe, E. (2011). Intelligent navigation systems for building evacuation. In Erol Gelenbe, Ricardo Lent, and Georgia Sakellari (eds.), Computer and Information Sciences II (pp. 339—345). Springer, London.

Guillemin, A., & Morel, N. (2001). An innovative lighting controller integrated in a self-adaptive building control system. Energy and Buildings, 33(5), 477—487.

Hardin, M. (2018). Post-occupancy testing of thermal dynamics of design-build residences in Tucson, Arizona. Technology Architecture+ Design, 2(2), 147—159. He, F., & Ma, C. (2010). Modeling greenhouse air humidity by means of artificial neural network and principal component analysis. Computers and Electronics in Agriculture, 71, S19—S23. https://doi. org/10.1016/j.compag.2009.07.011

Hiromoto, J. (2015). Architect & Design Sustainable Design Leaders. Post Occupancy Evaluation Survey Report. SOM: New York. survey-report_update_2.pdf

Jara, A. J., Genoud, D., & Bocchi, Y. (2014, May). Big data in smart cities: From poisson to human dynamics. In 2014 28th International Conference on Advanced Information Networking and Applications Workshops (pp. 785-790). IEEE.

Kazarian, A., Teslyuk, V., Tsmots, I., & Mashevska, M. (2017, February). Units and structure of automated “smart” house control system using machine learning algorithms. In 2017 14th International

Conference the Experience of Designing and Application of CAD Systems in Microelectronics (CADSM) (ррГ 364-366). IEEE. Kortuem, G., Kawsar, F., Sundramoorthy, V., & Fitton, D. (2009). Smart objects as building blocks for the internet of things. IEEE Internet Computing, /4(1), 44—51.

Klippel, A., Freksa, C., & Winter, S. (2006). You-are-here maps in emergencies—the danger of getting lost .Journal of Spatial Science, 5/(1), 117—131. Kobes, M., Helsloot, I., De Vries, 13., & Post,J. G. (2010). Building safety and human behaviour in fire: A literature review. Fire Safety Journal, 45(1), 1—11. https://doi.Org/10.1016/j.firesaf.2009.08.005 Li, N., Calls, G., & Becerik-Gerber, B. (2012). Measuring and monitoring occupancy with an RFID based system for demand-driven HVAC operations. Automation in Construction, 24, 89—99.

Marble, S. (2018). Everything that can be measured will be measured. Technology | Architectures Design,

2(2), 127-129. Nagy, D., Villaggi, L., Stoddart, J., & Benjamin, D. (2017). The buzz metric: A graph-based method for quantifying productive congestion in generative space planning for architecture. Technology|

Architecture+ Design, 1(2), 186-195. Park, D. (2017). Architecture from the bottom-up. Technology Architectures Design, 1(2), 140—142.

Patil, S. L., Tantau, H. J., & Salokhe, V. M. (2008). Modelling of tropical greenhouse temperature by auto regressive and neural network models. Biosystems Engineering, 99(3), 423—431. https://doi. org/10.1016/j.biosystemseng.2007.11.009

Pena, R., Meek, C., & Davis, D. (2017). The Bullitt Center: A comparative analysis between simulated and operational performance. Technology I Architectures Desion, 1(2), 163—173. https://doi.orc/10.10 80/24751448.2017.1354611

Pedersen, T. H., Nielsen, K. U., & Petersen, S. (2017). Method for room occupancy detection based on trajectory of indoor climate sensor data. Building and Environment, 115, 147—156. https://doi. org/10.1016/j.buildenv.2017.01.023

Phelan, N., Davis, D., & Anderson, C. (2017, May). Evaluating architectural layouts with neural networks. In Proceedings of the Symposium on Simulation for Architecture and Urban Design (pp. 1—7), Toronto, Canada.

Schwartz, M. (2013). Collaborative and human based performance analysis. In eCAADe Proceedings of the 31st International Conference on Education and Research in Computer Aided Architectural Design in Europe (pp. 365-373), Delft, The Netherlands.

Schwartz, M., & Das, S. (2019). Interpreting non-flat surfaces for walkability analysis. In Proceedings of the Symposium on Simulation for Architecture & Urban Design, Society for Computer Simulation International (pp. 287—294), Atlanta, Georgia.

Shelden, D. (2018). Cyber-physical systems and the built environment. Technology Architectures Design, 2(2), 137-139. Siemens, Building Automation — Impact on Energy Efficiency (p. 10), (accessed July 30, 2020)

Simondetti, A., & Birch, D. (2017). Computational design synergy: Stimulation through simulation.

Technology Architectures Design, 1(2), 143-145. Tang, S., Shelden, D. R., Eastman, С. M., Pishdad-Bozorgi, P., & Gao, X. (2019). A review of building information modeling (BIM) and the internet of things (IoT) devices integration: Present status and future trends. Automation in Construction, 101, 127—139. https://doi.Org/10.1016/j. autcon.2019.01.020

Thomas, B., & Soleimani-Mohseni, M. (2007). Artificial neural network models for indoor temperature prediction: Investigations in two buildings. Neural Computing and Applications, 16( 1), 81—89.

UcityLabAdmin (2019) SmartSantander: An unique city-scale platform. University Action Lab. https:// (accessed July 30, 2020)

Weiser, M. (1991). The computer for the twenty-first century. Scientific American, 265(3) September Issue, 94—100.

Wurzer, G., Ausserer M., Hinneberg H., Illera C., Rosie A. (2011) Sensitivity visualization of circulation under congestion and blockage. In Peacock, R., Kuligowski, E., & Averill, J. (eds.) Pedestrian and Evacuation Dynamics. Springer, Boston, MA. https://doi. org/10.1007/978-1-4419-9725-8_96

Zarzycki, A. (2016). Adaptive designs with distributed intelligent systems-building design applications. In Herneoja, A., Osterlund, T., & Markkanen, R (eds.), Complexity & Simplicity — In Proceedings of the 34th eCAADe Conference - Volume 1, University of Oulu, Oulu, Finland, 22—26 August 2016, pp. 681-690.

Zarzycki, A. (2018). Strategies for the integration of smart technologies into buildings and construction assemblies. In Proceedings of eCAADe 2018 Conference (pp. 631—640), Lodz, Poland.

Zarzycki, A., & Decker, M. (2019). Climate-adaptive buildings: Systems and materials. International Journal of Architectural Computings /7(2), 166—184. Zhao, H. X., & Magoules, F. (2012). A review on the prediction of building energy consumption. Renewable and Sustainable Energy Reviews, 16(b), 3586—3592. https://doi.Org/10.1016/j.rser.2012.02.049 Zuraimi, M. S., Pantazaras, A., Chaturvedi, K. A., Yang, J. J., Tham, K. W., & Lee, S. E. (2017). Predicting occupancy counts using physical and statistical Co2-based modeling methodologies. Building and Environment, 123, 517—528.

< Prev   CONTENTS   Source   Next >