Claims:One embodiment of the invention entitle is “Raspberry Pi Based Elderly Fall Detection System for Elderly People using IoT and Big Data”, we claim that:
1. Combining gyroscope and accelerometer data from both devices, this technique purposes to reduce false positives during detecting falls.
2. Our work focuses mainly on distributing resources in an IoT-based healthcare system and increasing the number of communications between smart devices and the cloud.
3. Decision-tree-based systems are gaining approval and are probably the best method for improving the accuracy and accuracy of fall detection.
4. This proposed fall detection solution based on Raspberry Pi, in which we used the accelerometer sensor to detect fall and navigation device to quickly locate a fall accident. The accelerometer sensor is attached to the Raspberry Pi; this system is designed specifically for indoor and outdoor environments and less costly than other solutions.
, Description:FIELD OF INVENTION

The growth of population in many countries in prerequisite of healthcare and to reduce the mobility in several countries demonstrates need for creation of assistive technologies to cater for public, particularly if suppose they need home-based treatment later being discharged from hospital. Thus, interactive submissions are often integrated within intelligent atmospheres on mobile devices. These systems typically have only limited properties and it is not skilled for processing vast capacities of data that discarded a lot of energy as devices are connecting to a cloud. Healthcare system practices data to improve the output in a micro fog based on IoT, provided that resources and enlightening health screening. The principal task of the invented system is to deliver low latency high data processing for resource-limited environment.

BACKGROUND OF INVENTION

In this modern age fall is a major threat to elderly people, falls are the source of unintentional injury death of elderly people according to the Centres for Disease Control and Avoidance Centres. The fall is described by the World Health Organization (WHO) as "an incident that leads to an inadvertent person is rest on the floor or at another lower level". Dropping on the floor is the world's second foremost cause of accidental damage death reported by WHO has fatal outcome around 0.42 million people fall. As of 2000, life expectation is raised by rate of five years owing to developments in medical field. The present-day population of ageing people (8.5 percent) will raise by 2050 depends on World Health Organization (WHO), representing 20 % of the world's population.

In particular, ensuring the elderly people's active and healthy aging (AHA) is one of the extreme challenges which also have a excessive opportunity for society in the upcoming decades. AHA's conception has recently categorized as a broad perception aimed at improving Quality of Life (QoL) of ageing as their age got increases, optimizing openings for health, contribution and safety. In this context, elderly people's health issues have become more severe and falls are the most frequent injuries, the extent of which can also require medical attention. Different non-computer-vision based technique for fall detection was introduced using different devices such as acoustics, floor vibration, acceleration, etc. to produce proof of sound, tremor and body movement to determine a fall.

There are many approaches to fall detection that support ageing, especially for those living alone. These systems are divided into three key types based upon sensor technology used: Wearable Based Systems (WS), Non-Wearable Based Systems (NWS), and Fusion or hybrid-based Systems (FS).

In particular, accelerometers are increasingly presence used in WS systems for the reason that they offer benefits such as consumption power is low, cost is very less, less weight, ease of operation, lesser in size, it can be placed on the different body localities and, most outstandingly, portability.

The performance of the device is not the same for all types of smart phones with different types of accelerometers and the sensitivity of the fall detection system often performed by calls and messages. Drop detection can also be tracked by smart watches and Wristband, but this device is not compatible because the user moves the hand at different speeds at any time, so the machine can not differentiate and drop from the daily life operation. A WS system depends upon machine learning (ML) methods have recently proposed to overcome the limitations and to improve accuracy for detection of falls. Machine learning method also used for fall detection such as neural network, vector support machine, dynamic Bayesian network model etc. to differentiate high accuracy fall event and daily life event but these systems are more costly. ML is a computer science technique which involves statistical inference from data models to make automatic predictions. ML constructs a model for predicting or to solve the given problem from training data. We are focussing on using several ML methods for fall detection in several works.

Raspberry Pi powered fall detection solution, in which we used the accelerometer sensor to detect the fall and navigation system to quickly locate the fall accident spot. The accelerometer sensor is attached to the Raspberry Pi, this system is designed specifically for indoor and outdoor environments and less costly than other solutions.

DETAILED DESCRIPTIONS

The FD system consists of a wearable device that is a Smart IoT gateway, a wireless communication network and cloud services, Raspberry Pi, accelerometer monitor, GPS module, GSM module and buzzer. Each element plays a key role for detecting falls. The raspberry pi is used as a central controller which controls and manage device aimed at fall detection. Accelerometer sensor is mainly for collecting data, while remaining modules such as GPS, GSM, and Buzzer are used in device output.

A. Accelerometer

This system incorporates the ADXL 345 accelerometer sensor, which can detect falling in X, Y and Z-axis. This fall detection sensors wear on the body, they monitor acceleration changes in three directions in moving this sensors, then the data analyzed to assess fall accident based on the stated threshold. The ADXL 345 accelerometer sensors sensing the acceleration rely on the single-axis, two-axis and three-axis function. This sensor has multiple features such as ± (2-16) g range measurement, motion detection, 4 mg/LSB sensitivity, 13 bit resolution, and flexible system interrupt. The above features make the ADXL 345 accelerometer sensor special for fall detection and reduce the algorithm's complexity with a small requirement for real acceleration values to use.

B. Raspberry Pi, GPS, GPS

Raspberry Pi is a mini single board computer, developed by Eben Upton in 2012 especially for experimental and educational purposes. Raspberry Pi's main stuff is to build low cost tool to enhance comprehension of hardware and programming skills in high-level analysis. Raspberry Pi device is used as control module in the proposed work.

GPS module communicates with 24 Earth orbit satellite-based navigation system providing position, timing, and velocity information. The GPS system comprises satellites, receiver and ground stations, where the receiver measures the distance from four satellites or more. The use of GPS16MU2 module in this invention is to track the location of the accident in fall.

ETSI (European Institute for Telecommunications Standards) establishes a standard Global Mobile Communication System. Low power is needed in the operation of this module which makes it suitable to communicate with low power consumption devices such as Raspberry Pi and link the Raspberry Pi to the mobile network worldwide. In GSM, we used the GSM SIM900A module to transmit data in the communication component that operates on the 900-1800 MHz dual-band and is more suitable for electronic projects by using serial port.

We use the ADXL 345 accelerometer sensor in the proposed system as an input device connected to Raspberry pi, which senses X, Y and Z-axis at a standard rate of 50 Hz due to the slow movement of elderly people. The findings analyzed show that 50 Hz sampling is batter then 100Hz sampling, because the precision and efficiency of the system depends directly on the algorithm of detection. The detection technique relies on the accelerometer sensor used on the human body, the accelerometer sensor uses acceleration with the aid of algorithm-based mechanism to detect fall and decision. Raspberry pi is used to extract the raw data from the accelerometer sensor and then compare it with the threshold defined. If the accelerometer value is greater than the threshold value the mean fall occurs. If a fall accident occurs, then Raspberry pi is responsible for allowing its Global Position System (GPS) to detect the location where the fall occurs, or to locate the location of the fall. Once the system finds the location for the fall detection, its GEO graphic coordinates were calculated and these coordinate and alert messages were sent to the responsible person via GSM massaging module and turned ON the buzzer.

C. Wearable Device

A model of wearable device was built by the mixture of three modular blocks: NUCLEO-L152RE, connected through one expansion board (X-NUCLEO-IDS01A5) along with 868 or 915 MHz sub-1GHz RF communication and an expansion board (X-NUCLEO-IKS01A1) established by ST Microelectronics. The NUCLEO-L152RE is well-found with a 32-bit ARM Cortex-M3 processor considered to deliver low power digital signal processing, very high performance and low voltage operation. Several tiny-ultralow-power sensors form the sensor board. Nevertheless, MEMS motion sensor (LSM6DS0) is utilized for collecting the motion data which occurs when adult performs ADLs or drops. LSM6DS0 is a 3D-axis accelerometer that functions at a complete range of acceleration (±2/±4/±8g). The software for wearable devices is depends on 'Contiki' an open source operating system (OS) built for restricted networks. By this the Contiki OS, we can receive full IoT stack support for 6LowPAN, that is, support for 6lowPAN, RPL (IPv6 Low Power Routing Protocol and Lossy Networks) and CoAP (through Erbium). Whereas Erbium is a low-power REST engine written in C language which offers Restful access for wearable device resources.

D. Wireless Communication Network

The low-power wireless IPv6 (6LowPAN) technology depends upon the IEEE 802.15.4 standard establishes wireless communication among the devices with the Smart IoT gateway. The 6LoWPAN is a tool intended to support the interoperability, synchronization and reliability of heterogeneous WSNs at lowest price comparing with other technologies such as Bluetooth / Wi-Fi and with very low demands. Moreover, this technology has essential benefits; greater address space, greater mobility, easy deployment and maintenance, which made this technology appropriate for use in IoT-enabled devices, particularly in resource-constrained devices. We develop and implement a 6LowPAN network consisting of two 6lowPAN nodes: a 6LowPAN border router (6LoBR) and a wearable device. 6LoBR plays a vital role when it comes to communication within and outside our 6LowPAN network. The 6LoBR is in charge for (i) sharing data among wearable devices and cloud services, and (ii) providing room for progressing and routing within the 6LoWPAN network. In this invention 6LoBR plays role of the Smart IoT gateway.

E. Emergency Alerts Handler

Use an MQTT-broker to send alerts for fall occurrence along with GPS location of the elderly person's house to the organizations answerable for caring the older people already enrolled in the program. Figure 4 illustrate a model of submitted notifications. It also sends that information to the cloud services. MQTT is chosen for being in substantial and stable IoT protocol. The MQTT offers SSL-based, end-to-end reliability and secure communication. It also integrates many levels of provision quality to validate processing of messages, from a non-optimal minimum level (QoS0) to a level of double recognition (QoS2).Because the network is very closely linked with healthcare of elderly, the QoS 2 service excellence standard has been designed to guarantee the efficiency of delivery of messages.

F. Performance Measures

Processing is done straight on the remote server in the context, without going by an intermediary environment which could process the applications. It could be noticed that the device performance improved after Claim Acceleration was observed while decreasing Acceleration was observed while sitting Acceleration was observed while walking the introduction of a micro fog as an intermediate layer, as resources are allocated with entire infrastructure. Both tests are carried out in actual surroundings. Therefore, Raspberry Pi 2 was used as a micro fog (i.e., an edge node to the network) and a remote server desktop. The measurements and parameters are selected for each scenario, along with results obtained, is presented in the following.

SUMMARY OF INVENTION

The Internet of Things is a new standard that helps the adult population in advance their quality of life by promoting a more centralized and personalized model for treatment. This proposed technology is fall detection system using three axis accelerometers by Raspberry PI. The introduced innovation FD-system, an IoT system for the identification of elderly people depends upon a Big Data model which uses decision trees-based ML processing methods. The detection algorithm works on threshold value if the input value is grater then defined threshold then the device monitors the person's position, sends warning message and location to the responsible person's sand switch on the buzzer via GSM module. In order to predict the dropping of people, the FD system proceeds the acceleration measured in x, y and z axes by the movements of the elderly as data, that were obtained with a 3D-axis accelerometer sensor mounted in a wearable device centered on 6LowPAN.The tests checked which were carefully performed by four volunteers, showed that the accuracy of our proposed fall detection system could exceed 99.9 per cent in fall detection. The tool was put on the waist of the elderly, and it provides appropriate solution for use in an indoor location for any elderly person. The model gives remotely alerts healthcare professionals, emergency centres, caregivers, and the family members of the elderly in the event of a fall using QoS appliances.
5. DESCRIPTION OF THE DRAWINGS
The detailed description should refer to the following drawings in which they refer to elements like numerals and in which they refer to:

Figure 1. Architecture Flow of Fall Detection
Figure 2. Algorithm Flow of Fall Detection
Figure 3. Fall Detection Architecture Flow