Research On Indoor Geo-fence Based On Mobile Phone Sensor

Geo-fencing plays a very important role in the areas of information push, smart home,attendance sign-up,child safety monitoring and intelligent medical care. The core of the Geo-fencing is positioning technology, and GPS-dominated outdoor positioning research has been relatively mature, but in the indoor, underground passage,or other complex and volatile environment, it could not provide high-precision continuous navigation through GPS. Large-scale popular mobile terminal has become the most widely used navigation terminal equipment, it is a good prospect to study the indoor navigation technology based on the mobile terminal sensor and apply it in the field of indoor geo-fence.However, there are challenges of realizing the indoor geo-fence based on the mobile terminal sensor: the mobile terminal sensor has low sampling rate and high noise. The mobile terminal has a high degree of freedom and high degree of freedom of movement. The direction of the mobile terminal is not equivalent to the pedestrian heading. The performance is limited within mobile. Indoor virtual geographical fence requires high navigation accuracy Firstly, we analyzes the advantages and disadvantages of various positioning technologies, taking full account of the challenge of the local fence of the mobile terminal sensor. we adopts the pedestrian route estimation method and applies it to the indoor geo-fence model. Based on pedestrian movement physiological characteristics we analyze the characteristics of the mobile terminal sensor signal , and calculate the pace, step and heading in a variety of application scenarios and calculate pedestrians in the indoor plane displacement, and monitor whether pedestrian has left the virtual geographical fence area; Secondly, as it is not suitable for pedestrian mobile terminal with complex pose and movement in indoor complicated environment. we study the pedestrian mobile terminal in the interior of the survey room it could not be applied to the traditional pedestrian route heading algorithm, such as initial calibration, error dynamic compensation, error model under sub-stable scene, we analysis these characteristics of the mobile terminal direction sensor signal by wavelet transform, and uses the neural network semi-supervised learning model to predict the user's heading angle method in various use scenarios.The experimental results show that the indoor geo-fence solution based on the mobile terminal sensor is feasible. When the user in the indoor virtual fence area activities in the length of 15 to 60 minutes, the geographical fence real-time accuracy rate of 93.5%. When the user event is activated after 3 seconds, The accuracy rate of the geographical fence reached 98%, and the improved accuracy of the heading forecast reached 96.6%.