Indoor Positioning And Smart Sensing By Exploiting Visible Light

Flourishing of mobile computing and rapid development of smart-device change the whole life style of mankind.The mobile smart-devices and the widely deployed in-ternet of things(IOT)infrastructure not only enable information processing at anywhere and anytime,but also allow sensing of the device-holder and the surroundings.After understand 1)the location and 2)the activity of the user,the smart-devices are able to enlarge the limited sensing range of an individual mankind.To achieve this goal,the industry society and the research society propose many positioning and sensing systems to satisfy the ever-increasing demand.However,for the positioning systems,the current mainstream WiFi-based indoor positioning systems are vulnerable when facing with the environmental changes.In the meantime,the deficiency of the signal source limits the coverage area and influences the precision of the system.For the sensing systems,the mainstream computer-vision techniques suffer due to the high deployment cost and the privacy issues caused by the high resolution photos.Visible light is the most ubiquitous electromagnetic wave in our daily life,and both smart-devices and human-eyes can per-ceive its appearance.Visible light is of high directivity and resists to the environmental changes.In the meantime,the densely deployed illumination infrastructure in indoor environment can directly be used as the signal source,which reduces the deployment cost and the complexity of the positioning/sensing system.The intriguing characteristic of visible light makes it possess a high potential to be the medium of positioning/sensing in the next generation.However,it is challenging to exploit visible light for positioning and sensing.First,different from WiFi signal,the received light strength(RLS)is just a scalar,which lacks the unique identity that can differentiate the light sources.Second,different from WiFi signal,RLS is extremely sensitive to orientation and altitude of the receiver.As the consequence,the variable space of RLS is of a high dimension,thus makes it difficult to build the relationship between the RLS set and positions.In addi-tion,it is a challenging task to recover fine-grained user-activities from a coarse-grained RLS set,due to the high degree of freedom(DoF)of some user-activities(e.g.,gesture of hand).Facing with these challenges,the major research contents and the innovations are as follows:· Designment of a visible light based positioning system,Lightitude.The key ob-servation of Lightitude is that,the RLSes perceived in different indoor locations vary,and this difference is obvious for the light sensors.Thus,RLS can associate with different indoor positions.Lightitude first proposes and validates a light strength model to describe the relationship between the RLS and the status of the device,then designs a particle filter to locate the user.Evaluations in typical indoor environments confirm the effectiveness and the robustness of Lightitude.Lightitude performs well even with the shading effect of obstacles,unpredictable behaviors of users,interference of sunlight and human-body's shading effect.·Designment of a visible light based indoor recommending system,LiLoc.The key observation of LiLoc is that,different pathways possess their own RLS char-acteristics.As the consequence,the service provider can establish the information database with almost zero-cost,by conducting only a once-for-good traverse in indoor environment.Leveraging the nature mobility of users in big indoor en-vironments,Liloc System first locates the user by checking the database,then pushes the recommending information to the users according to their locations.In addition,by excavating the fusion of sensor data,LiLoc unveils three typical user-behaviors that demonstrate the user's potential preference,thereby facilitates a complementary recommending after users' departure.Experimental results val-idate the effectiveness of LiLoc in real-world scenarios.· Designment of a visible light based hand sensing system,HandSense.The key idea is that,the human hand can act both as an opaque obstacle and a reflector of light.As an opaque object,human hand not only blocks the line-of-sight(LoS)path from light sources(e.g.,sunlight,LEDs)to light receivers,but also reflects the visible light,resulting in a weak backscattered light signal.These effects are recognizable to light receivers.As the consequence,a specified hand status corresponds to a set of specified RLS.Our researchers implement the blocking detection and the backscatter detection on the server.In addition,by integrat-ing the inherent constraint proposed by the hand anthropometric model,our re-searchers design a heuristic optimization algorithm to realize the hand sensing system.Evaluations in real-world settings confirm that HandSense system can precisely sense the silhouette,movement and even the gesture of the human hand.