In recent years, indoor air quality (IAQ) has drawn considerable attention in both the public and scientific domains. This is mainly due to two reasons. First, people spend a majority of their time indoors, for example, about 90 percent for people in the United States. Second, most buildings appear to fall far short of reasonable air-quality goals (Huizenga et al. 2006). Statistics from the U.S. Environmental Protection Agency (EPA) indicate that, on average, the indoor levels of pollutants are two to five times higher than outdoor levels (U.S. Environmental Protection Agency Green Building Workgroup 2009). Bad indoor air quality influences human health, safety, productivity, and comfort (Wyon 2004; Daisey, Angell, and Apte 2003). Personal exposure to air pollutants is highly variable due to the presence of indoor air-pollution sources. Providing personalized IAQ information has the potential to increase public awareness of the relationship between people's behavior and air quality; help people to improve their living environments; and also provide valuable information to building managers, policy makers, health professionals, and scientific researchers.
IAQ monitoring is challenging because indoor air-pollutant concentrations and human motion patterns each vary spatially and temporally within and across rooms. These variations are caused by differences in user activities, number of occupants, and ventilation settings. For example, two office rooms in the same building may have significantly different IAQ because of variation in the number of occupants (Huizenga et al. 2006). Existing solutions that require stationary sensors or target mobile outdoor sensing scenarios are inappropriate for personalized IAQ monitoring. Stationary sensing (Godwin and Batterman 2007) has several limitations. First, it can only measure the IAQ experienced by those who happen to be near the sensors and there can be substantial variation in IAQ even within one room. Second, when locations or rooms outnumber people, achieving full occupant coverage with stationary sensors is more expensive than doing so with personal mobile sensors.
A mobile sensing system designed for personalized IAQ monitoring must meet the following three requirements. First, the system requires accurate and reliable indoor localization techniques to detect user locations. Outdoor mobile sensing solutions use GPS localization, which fails indoors. Some existing approaches use proprietary radio frequency and ultrasound technologies for room localization, which require investment in infrastructure and special hardware worn by all users. Others use Wi-Fi-based fingerprinting, which requires time-consuming precharacterization and is hampered by device or environment heterogeneity. Second, the mobile sensing devices must be inexpensive and portable. This limits the number and types of sensors that can be integrated within each mobile device. Existing air-quality sensing solutions require multiple types of sensors, each of which covers a subset of pollutants. This can be prohibitively expensive for personalized mobile IAQ sensing. Achieving high-quality IAQ monitoring with inexpensive sensors is challenging. Third, the system needs to achieve a good balance between energy consumption and coverage. IAQ sensing depends largely on the motion patterns of individual users. This leads to redundant IAQ information when users are near each other and may lead to gaps in coverage for users who are not presently carrying sensing devices.
This article describes MAQS, (1) a personalized mobile sensing system for IAQ monitoring. MAQS estimates anthropogenic air-quality factors (for example, C[O.sub.2] and contagious viruses) using C[O.sub.2] concentration and estimates other air-quality factors (for example, volatile organic compounds [VOCs]) using air exchange rates. MAQS integrates smartphones and portable sensing devices to deliver personalized, energy-efficient, IAQ information. …