*2.3. Location-Enhanced HAR*

The concept behind location-enhanced activity recognition is to use the location of a person as a feature of an activity. It is reasonable to assume that certain activities can only be performed in certain areas or locations. For example, in a home setting, food preparation would take place in the kitchen, while brushing your teeth would be performed in the bathroom. A recent survey conducted in [30] highlights how the location characteristic, as well as other characteristics (e.g., time, conditions, duration) of an activity can aid in the living of elderly people. Moreover, in [31], the authors show how having these additional characteristics can result in enriching activity modelling and recognition in providing assisted living in smart homes, resulting in activity classification estimates ranging from 88.26 to 100% for basic activities. Furthermore, gaining knowledge of a persons location can be used as an alternative method of improving activity classification as shown by [32], where the authors concluded that adding location awareness aides in activity recognition. Finally, the use of location-enhanced activity recognition grants the benefit of being a more unobtrusive approach as highlighted by [33] than other approaches that use more on-body sensors, as discussed in Section 2.2.



**Table 1.** HAR publication details.

*Legend***.** IMU sensors: Acc, Accelerometer; Mag, Magnetometer; Gyro, Gyroscope; Classification approach: SVM, Support Vector Machines; KNN, k-Nearest Neighbours; LR, Logistic Regression; ANN, Artificial Neural Network; HMM, Hidden Markov Model; NB, Naive Bayes; DT, Decision Trees; RF - Random Forest; CRF, Conditional Random Field; COTS, Commercial-Off-The-Shelf.
