Falls have always been source of injuries and fatalities. There can be various reasons for falls from medical conditions to loss of balance. Fall detection in the labs is based on video recordings while ubiquitous methods utilize inertial sensors. These methods require a lot of data processing consequently much power is utilized. In this paper, we propose a fall detection method which does not require inertial sensing nodes thus reduces computation and power requirements. Previous methods have either used received signal strength as monitoring parameter which is quite unstable or uses ambient monitoring points sensing through which remains disturbed by environmental noise and obstacles. We are proposing a novel ubiquitous wireless nodes deployment on the subject's body to minimize noise interferences. Fall has been detected using wireless sensing physical layer channel state information (CSI). The human body acts as an obstacle for the wireless signals generating unique signature with respect to the activity performed which can be estimated using CSI. This CSI information is available in new 802.11n WLAN NICs. We argue that this signature should be best observed when both the sender and receiver nodes are deployed on the body due to minimum environmental interference. We have trained the system to identify these signatures and detect falls. We have also been able to differentiate falls from fall like activities such as squatting and sitting on chair with sufficient accuracy. © 2015 IEEE.