1 Introduction

Wireless sensor networks (WSNs) consist of internetworking multiple sensors, which are largely deployed on an open and remotely uncontrolled environment including multitasking capabilities [2]. These sensors lie scattered in an unattended environment (i.e., open field) deployed at remote locations which are situated far from user locations. Design of an elderly support mobile healthcare and monitoring system on WSN for dynamic monitoring of the elderly people was given in [15]. They can be invoked as automatic alarm for handling emergency situations as well as playing role of living assistant [16]. Personal health management was set up to permit health experts to view and update current and log reports of elderly people. The device sets thresholds for sensors and manages experts remotely, which is an essential part for any monitoring model of elderly people [4].

In [12], the authors analyzed quality-of-service (QoS) aspects related to implementation of WSN for supporting elderly healthcare and health assistant in patients’ health monitoring. Sweetser et al. [7] discuss signal localization algorithm collected from sensor data to monitor patient’s health in healthcare clinics. Lewandowski et al. [6] showed a WSN-based implemented healthcare application for analysis of healthcare monitoring systems. The WSN-based designs of sensor networks for applications such as support for health, collect heart activity data, and transmit data over 802.15.4 network nodes were discussed in [1]. Kumar et al. [15] suggested a related device hardware and operational software for a representative system, which supports research aspects towards time synchronization, power management, and chip signal processing. Based on sensory data obtained from clinical and ambulatory settings, the system can support towards understanding the limitations of device as well applications of the WSN technology based upon the in-depth study. Real-time monitoring of health with electrocardiogram (ECG) sensor was investigated in [18]. Lee and Song [13] measured data being transmitted over WSN network which involves remote monitoring and health support. Buratti et al. [19] focused on the design of microsubscription management systems with implementation of dynamic memory kernel over variable payload multiplexing related to service-based events that are the major outcome. It yields good results for e-health applications. Other works of support systems in WSNs can be seen in [3, 8,9,10,11, 14, 17, 20, 21, 25,26,27,28,29,30,31,32].

Patient diagnosis system contains the central care units for providing the constant display and understanding the patient’s health updates. The patient diagnosis system uses the latest communication technology for the treatment of the patient. This system provides the several measurement methodologies and the signal processing methods. The quality of this type of systems is measured using communication techniques and software technologies.

The objective of this research is to design and develop a health monitoring system that comprises compact body worn devices (embedded on a garment) to monitor and aggregate data from patients, as a network connected over web server and mobile network as shown in Fig. 1. The primary objectives of automated patient diagnosis (AUPA) are as follows:

  1. (i)

    To survey and analyze the existing approaches to provide consistent QoS aspects over service differentiable WSN networks.

  2. (ii)

    To support minimizing packet loss and delay using both real-time and simulated approaches.

  3. (iii)

    Handling erroneous data (aggregated data may have missing values or errors in certain packets). Accuracy of information sent to end user would be better.

  4. (iv)

    Providing optimized decision-making for healthcare support.

Fig. 1
figure 1

Stack architecture of AUPA

Full size image

The importance of creating alerts whenever the abnormal condition happens is addressed in our proposed work. The results suggest the problem of current emergency reply for the patient. The system is designed to reduce the possible critical events and global costs for healthcare systems.

A QoS architectural framework is established for the design of AUPA as healthcare and monitoring kit. It comes from the need for consistent monitoring and maintenance of healthcare among patients and elderly people that leads to the demand on change in traditional monitoring approaches among chronic disease patients and alert on acute events. The demand on need to capture transient abnormal events reliably and detecting life-threatening disorders normally go undetected since it happens only infrequently [22]. Stress on better diagnosis and support on treatment with monitoring are possible through communication systems enabled with WSNs [22]. Such approaches demonstrate better results for monitoring of patients over surgery, rehabilitation scenarios, and daily assessment of life activities among elderly people or people observed as suffering from cognitive disorders like Parkinson’s, Alzheimer, or similar diseases.

A case study for healthcare and gathering patient’s data is carried out to satisfy the following objectives:

  1. (i)

    To enable physicians or experts to consistently monitor specific health aspects of patients or elderly members through sensor-based device wearable kit.

  2. (ii)

    To gather patient’s data and develop knowledge repository by employing supervised learning approaches towards disease prediction and suggestive approaches.

  3. (iii)

    To integrate patient’s history and provide support through monitoring with effective diagnosis approaches.

2 Method

2.1 AUPA design

The design factors and challenges of design of AUPA over WSN hold on data depletion, QoS management, robustness over dynamic environment, and highly scalable to numerous sensor devices. Few additional suggestions to the primary challenges are summarized as follows:

  1. (i)

    Reduction of active traffic load intensity over multiple WSN devices involves data communication minimization over defined wireless channel which includes data aggregation and communication of the network state along with summary of actual data transmitted.

  2. (ii)

    Maximize the intensity of packet loss along with network life time and support towards achieving adaptive QoS.

  3. (iii)

    Scalability can be improved by organizing network in a hierarchical clustered approach which utilizes localized interaction among sensor devices over localized algorithms among sensor devices, at the same time robustness to changes in environment. The algorithms can be improved through self-learning and organization, which also includes self-healing and self-adaptive over the configuration setup.

The primary focus of AUPA kit is focused on data maintenance and handling effective decision-making among large networks. AUPA can be defined as a WSN MOTE with the following properties as suggested in Table 1.

Table 1 Properties of AUPA
Full size table

Two major supporting functionalities of AUPA are data aggregation [6] and data management factors for variable services along with healthcare-based decision-making. AUPA controls data gathering and analysis through functionality of interrupt handler, where the device collects external environmental parameters using health-supported sensors, process them through processing units, and support data management and decision-making through optimized swarm optimization algorithm such as Ant Colony Optimization (ACO) for analysis.

AUPA adopts the stack architecture structure where each device is equipped with sensory interface setup which can interface multiple sensors such as temperature, EKG, EMG, motion sensors, and body sweat and moisture which are highly sensitive to variable environments as shown in Fig. 2.

Fig. 2
figure 2

AUPA functionality model and flow diagram

Full size image

AUPA is developed using ATmega 328 microcontroller (see Table 2), which is a simple, cheap, and effective microcontroller for industry-specific applications. It can operate at a low voltage of 5 V for simple applications. It possesses maximum of 14 slots for carrying out multiple services and handles event-specific tasks based on actuations of users. The kit has been developed using C programming which captures multiple sensors or devices for interactivity. WSN nodes are heterogeneous; hence, multiple sensors are built along with the kit to actuate and partake in completing an event as per need of service in use. As ATmega possesses sufficient flash and programming memory, the kit handles large datasets to be stored in external memory and implement monitoring aspects of applications such as habituate monitoring, healthcare monitoring of patient, agricultural farm-specific environment, and pollution control.

Table 2 AUPA—components
Full size table

AUPA supports inbuilt wireless sensor motes due to its on-board sensor modules, including functionalities such as data logger, and wireless network connectivity for reliable session quality. AUPA works on IEEE 802.15.4 standard, on supportable USB interface through which device can support the data to be gathered. The AUPA equips 16-MHz ATmega 328 microcontroller with 1024-byte EEPROM memory and 64-KB liner programming flash memory as shown in Fig. 3. As optional accessories, AUPA includes sensor for data gathering and transmission including temperature, ECG, EKG, and sweat and moisture sensors as shown in Table 2.

Fig. 3
figure 3

AUPA designed as healthcare kit with sensors

Full size image

The trans-receiver adopts a data transmission rate of 256 kbps at 2.4 GHz whose transmission range is 50 m within an indoor environment and 100 m in outdoor scenario. The experimentally gathered data have to be temporally stored on the wireless sensor board before being collected and used for decision-making/analysis. AUPA can also be fixed on to 1-MB external flash, which stores the data and buffers the process. The experimental test bed is explained using the organizational setup and test procedures. The healthcare sensors which are used in this project are:

  1. (i)

    EMG sensor, which gathers electrical signals produced during contractions of muscle.

  2. (ii)

    EKG sensor, which collects electrical signals produced by the heart (ECG biosensors).

  3. (iii)

    Sweat and moisture sensors, to gather the body sweat and moisture level.

  4. (iv)

    Motion sensor, to gather the abnormal and normal motion of a person.

2.2 Architecture and functionality of AUPA

AUPA mote works on software which manages a fixed array of wireless sensor network nodes fitted with wireless interface backchannel boards allowing data logging and transmission of data over multiple nodes. AUPA mote sensor test bed supports data aggregation and addresses challenges through wireless interconnectivity as interface. The kit supports automated test bed programming and gathers log data generated by experiments into a PC or server as persistent database. The sensor data can be retrieved as source from database, where the database maintains consistent data collected at various iterations including sensor data, time, sensor name, and service adopted from test bed. AUPA is attached with different software components, such as:

  1. (a)

    Data logger and reading sensor data.

  2. (b)

    Wireless network interface for communication.

  3. (c)

    Port/USB interface for collecting the sensor data.

The architecture is based on a multi-tiered protocol structure, which relies on ZigBee (network and application layers) and 2.4 GHz, IEEE 802.15.4 (physical and data link layers) communication protocol standards, which serves as a network backbone for data transfer and routing phenomenon. To understand and analyze the behavior of WSN network, an application-based test bed is developed with the objectives of implementing, assessing, and validating AUPA architecture. Figure 4 demonstrates the overall design for AUPA which contains the wireless sensor networks, base station, and control unit. The control unit is connected to the Internet, emergency department, and also the mobile user. The Internet can communicate with medical specialists for further updates.

Fig. 4
figure 4

Overall design for AUPA

Full size image

2.3 QoS model architecture

AUPA adopts variable QoS modeling architecture, where multiple WSN nodes are under limited mobility. Any WSN node can act as a source or as destination or as an intermediate node for data transmission. The intermediate forwarding node as A or B … G or H can send or receive the data received from another node as shown in Fig. 5.

Fig. 5
figure 5

QoS support with route establishment

Full size image

AUPA works on multiple sensors such as temperature or humidity or ECG sensors with required buffer to “hold the data” or “transmit the data” to another AUPA node. The experimental test bed is explained using the organizational setup and test procedures. AUPA node referred to as “ni” adopts the following node configuration and route establishment process as shown in Fig. 4. We assume that there are N assorted sensor nodes arranged, which are formed by a connectivity graph of G = {V, E}, where E represents wireless connections between active nodes. For dissimilar types of active nodes, the broadcasting bandwidth and broadcast control are dissimilar. All connections are unspecified to be directional, and each is connected with a metric representing its connection superiority. To perform sensing responsibilities, there are M active nodes selected from the dynamic sources. All data packets should be launched to the sink node inside the necessary target, where dissimilar types of active nodes contain their individual target constraints. The intention function of the delivery assignments is that all data packets require to be transmitted with the least sum of individual cost. The duration of an active node in the network is defined as the occasion for it to reduce its power. For every Link, Linkxy belongs to E and each active route z, we mention Linkxyz as,

$$ {\mathrm{Link}}_{xy z}=1\ \mathrm{if}\ \mathrm{active}\ \mathrm{route}\ z\ {\mathrm{includes}\ \mathrm{Link}}_{xy} $$
$$ {\mathrm{Link}}_{xy z}=0\ \mathrm{if}\ \mathrm{active}\ \mathrm{route}\ z\ \mathrm{does}\ \mathrm{not}\ {\mathrm{include}\ \mathrm{Link}}_{xy} $$
$$ {\begin{array}{c}\ \\ {}A\end{array}}_{xy}=\frac{C-\mathrm{Minimum}\ \left({\mathrm{Link}}_x,{\mathrm{Link}}_y\right)}{{\mathrm{Quality}}_{xy}\ \mathrm{x}\ {\mathrm{Quality}}_{yx}}\ \mathrm{x}\ {\mathrm{Time}}_{xy}\ \mathrm{x}\ {t}_{xy} $$
(1)

2.3.1 Node configuration “ni”

Any node “ni” can be considered to be part of AUPA network, only if the node is accepted using the configuration procedure. An AUPA node is accepted as part of network, only when it is affixed with unique name value and static IP addressing format within its ROM.

2.3.2 Coordinator node

Any set of intermediate nodes can be defined as coordinator node (CN) nodes or coordinator nodes. Figure 4 shows nodes E and F as coordinator nodes. Nodes which can act as major buffering node can be taken as coordinator nodes. The coordinator nodes link route among intermediate nodes such that multiple source and receiver nodes can communicate. The coordinator nodes maintain variable buffer density such that multiple intermediate nodes do adopt upstream and downstream through coordinator nodes.

2.3.3 Route establishment process

When any AUPA node “ni” receives an instruction or message, it returns an acknowledgement to its sending or upstream node as shown in Algorithm 1.

Algorithm 1: AUPA_Centralized_Algorithm

figure h

On receiving this acknowledgement, a node can confirm its route update or link with its neighboring nodes. If the node “ni” does not respond to any instruction or message after many retries, its upstream or sending node will report the node failure to the coordinator node; then AUPA can diagnose the problem with any failure report.

2.3.4 AUPA QoS manager

The QoS manager focuses on providing an adaptive QoS supportive route establishment approach between source and receiver. Allocation of variable buffer among intermediate nodes determines the adaptive path for data transfer. Any intermediate node “ni” can participate in route determination along with coordinator node between the source and receiver node. Multiple sensors gather data and transfer the data between the nodes. The data are stored in intermediate buffer of nodes selected for transfer as explained in Algorithm 2.

Algorithm 2: AUPA_QoS_Manager

figure i

The node which participates in route determination updates its status at intermediate intervals of time of experimental runtime. The experimental test bed is explained using the organizational setup and test procedures.

2.3.5 AUPA reversed push path insertion

At the initial stage of reversed push path insertion, for every active node, the calculated path for minimum cost from the beginning node to the sink is established. Reversed push path insertion then discovers the active node that has the prevalent path cost to the sink node and augmentation selection of candidate nodes with the minimum supplementary insertion cost as shown in Algorithm 3.

Algorithm 3: AUPA_Reversed_Push_Path_Insertion

figure j

3 Results

The performance of AUPA’s QoS approach is tested using a set of health-related sensors which gather the patient’s data over variable period of time and send from a source to destination AUPA node. The experimental test bed is prepared as shown in Fig. 5, where the AUPA kit can be deployed on a ZigBee communication kit.

3.1 Test bed preparation

AUPA adopts a step-by-step design process and can be configured over a third-party software tool for autonomous network composition, configuration, and optimization. The AUPA sensor nodes are deployed in a rectangular room measuring 20 ft × 25 ft, where each node is configured over coordinator tools and deployed through gateway [5]. WSN experimental approach of QoS analysis depends upon data logging and data dissemination procedures along with routing program binary as shown in Fig. 6.

Fig. 6
figure 6

AUPA test bed setup

Full size image

3.2 Network programming

Interconnectivity between various AUPA nodes is made feasible using a configuration tool [6] which assigns a name for AUPA node being active. The program image for an AUPA node is written into program memory which gets activated when node is deployed in an experimental process.

3.3 Data gathering

The experimental data generated during the process in iteration is valuable for understanding the behavior of WSN nodes. The coordinator node also maintains the basic log and operations carried out by node. The log operations maintained also support in understanding the errors [16].

3.4 Accuracy in data collection (Table 3)

The experiment follows (i)–(vi) steps:

  1. (i).

    Aggregate sensor data

  2. (ii).

    Store data in variable buffer

  3. (iii).

    Analyze data on transmission

  4. (iv).

    Transmit data to another node

  5. (v).

    Maintain the interval

  6. (vi).

    Clear the buffer data

Table 3 Test bed generation
Full size table

The experimental process adopts the steps to collect the data, store, analyze route for proper route selection, and then transmit to another adaptive route based of buffer allocation and data collected. If the transmitted data are not received by intermediate node or co-coordinator node, then re-transmit is requested between sender and receiver node until step (v).

The signal strength of the received data in the network is called the accuracy. It is based on the network setup, route maintenance, and how the sensor data is aggregated.

3.5 Data aggregation and dissemination

The experiment is carried out based on collected data set size and transmission size. Multiple test iterations are adopted to understand the QoS behavior of AUPA nodes under variable environmental situations of the patient. Patient data are collected from the sender to the receiver, while QoS metrics such as packet loss, end-to-end delay, and other congestion metrics.

3.6 Network deployment and connectivity issues

The AUPA nodes in network are monitored based on signal link capacity while at runtime is analyzed based on the multicomponent optimizer system implemented on sensor gateways. The state of the network at runtime can be observed on a device. The internode connectivity establishment time observed based on packet size, node power consumption, and observed delay for each delivery can be understood from data readings. The deployment strategies followed in solving coverage problem for AUPA are manually assigned initially and then updated at random intervals based on its AUPA node signaling capacity. The possible strategies to be adopted can be force-based, grid-based, or computational geometry procedure-based.

3.7 Indoor setup

AUPA test bed creation to define QoS and efficient route control approach is arranged indoor within a room of 10 × 15 m as shown in Fig. 7. AUPA nodes are placed with an interspacing of 5 to 10 m intermittently. The naming of nodes indicates that node “A” is defined as source node while node “D” as receiver node.

Fig. 7
figure 7

Indoor test bed implementation

Full size image

Nodes communicate with each other and possess limited mobility as a patient lying on bed or moving within a room, but not outside the campus or moving into another room. Each node establishes a communication link with the coordinator node which creates the adaptive route for communication. The sensor data transmitted over the network generate traffic over the route created with packet streams whose mean packet size possesses 450 bytes (including TCP/UDP, IP headers). The data transmission generates an approximation of 3500 to 5000 bytes of data traffic over multi-nodal network.

The back-end programming is used to analyze the data which is produced by the sensor nodes and providing the data communication in the AUPA architecture for implementing the indoor setup.

3.8 Main results (Figs. 8, 9, 10, 11, 12, 13, and 14)

Fig. 8
figure 8

Observed delay of AUPA

Full size image
Fig. 9
figure 9

Measured packet loss over AUPA, LEACH, and SPIN route approaches

Full size image
Fig. 10
figure 10

Traffic load analysis for AUPA approach over data transfer

Full size image
Fig. 11
figure 11

Reachability over AUPA and existing approaches

Full size image
Fig. 12
figure 12

Packet overhead of AUPA with existing approaches

Full size image
Fig. 13
figure 13

Latency for AUPA with existing approaches

Full size image
Fig. 14
figure 14

Jitter for AUPA with existing approaches

Full size image

4 Discussion

The performance analysis of AUPA routing scheme is discussed in this section. The experimental test bed uses 10 WSN nodes implemented over limited random mobility speed and conserved over variable real-time data transmitted between WSN nodes. Behavior of AUPA routing approach and its performance is observed over QoS metrics such as throughput, traffic load intensity, observed end-to-end delay, and effective signal connectivity range.

AUPA scheme demonstrates highly reduced signal loss during session in use based on optimal route selection and establishment, where the beacon signal is also considered as a QoS metric. Compared with existing schemes, namely SPIN [23] and LEACH [24], AUPA outperforms with minimal signal loss rate and hence a better neighborhood node selection and link selection is guaranteed. Figure 8 denotes the observed signal rate and loss rate when numerous WSN nodes participate in route establishment and communication. It can be understood from the observed delay rate that AUPA maintains a nearly average minimal delay compared to SPIN and LEACH route schemes, due to adaptive node selection in route establishment. This work does not consider end-to-end delay due to route establishment time due to intermediate node availability.

Figure 9 explains the measured packet loss achieved over 10 WSN nodes possessing AUPA properties. It can be noticed that AUPA showcases an average of 45.62% of packet loss while SPIN shows an average of 64.75% of packet loss. An abnormal behavior is noticed when both SPIN and AUPA show 46% of loss when 5 WSN nodes are used in experiment, which is primarily due to sleep period of WSN node. SPIN shows an abnormal activity, while AUPA demonstrates an expected loss rate.

Figure 10 demonstrates traffic load intensity of three different schemes, namely SPIN, LEACH compared with AUPA. AUPA shows an average of 57.24% of traffic intensity observed over 472 ms of experimental run, compared to 538 ms of LEACH and 594 ms of SPIN routing schemes. AUPA was able to demonstrate a minimal time due to selection of node with better buffer capacity for coordinating in routing process.

Figure 11 explains the readability over WSN nodes possessing AUPA can be identified in linear time. AUPA shows an average of 88.75% of linear time of experimental run compared to existing approaches. The extent to which a node in graph is reachable from others in the proposed work will be used for the quality of service. Figure 12 denotes the observed time it takes to transmit data on a packet-switched network. Figure 12 shows that the AUPA approach of each packet requires extra bytes of format information that is stored in the packet header. It reduces the overall transmission speed of the raw data.

Figure 13 shows that latency is an expression of how much time it takes for a packet of data to get from one designated to another. Figure 14 demonstrates that jitter in the scheme is inversely comparative to bandwidth and as the bandwidth of the scheme enlarges, the jitter also diminishes. The proposed AUPA method diminishes the jitter compared to the related algorithms such as LEACH and SPIN.

To sum up, the proposed AUPA shows better performance than the related methods (SPIN and LEACH) in terms of delay, packet loss, traffic load, reachability, packet overhead, latency, and jitter.

Distributed architecture maintains the data interoperability for the interaction within dissimilar data application. This interoperability will increase system balance and accessibility. AUPA architecture embraces distribution as a significant central process. AUPA architecture is having high-quality application programming interface and product maintenance. Architecture authorizes the reusability and litheness of the product. In addition to that, architecture is most vibrant and application nodes are included at runtime. The architecture drivers are mainly designed for improving the scalability and accessibility. In cloud computing environment, the distributed services are more eminent in near future. Data interoperability is the building of the independent interface at a significant time frame.

AUPA is a distributed architecture design strategy for automated patient diagnosis. AUPA outperforms the existing schemes, namely SPIN and LEACH, with minimal signal loss rate and an enhanced neighborhood node assortment and link selection. It diminishes the jitter compared to the related algorithms.

5 Conclusion

The primary objective of AUPA is to support effective QoS provisioning of WSNs, where providing healthcare for patients who require consistent monitoring and diagnosis. The architecture and implementation provide the design strategy of WSN for healthcare monitoring system based on the following society and environmental aspects.

Social impact

This work has a major social impact of supporting the patients in hospital and providing a consistent monitoring of physically disabled people who are in need of support related to medical needs. AUPA kit can help to minimize chances of criticality that arises among patients in case of any emergency. Most of elderly people do not have support for immediate medical care due to hospitals available at remote locations and immediate availability of expert physicians in nearby locations. Hence, such kits do have major social impact among patients and elderly people.

Economic impact

The development and maintenance cost of kit is less due to its simplicity. AUPA kit can be easily deployed on the cloth of any human being or can be worn on the wrist such that health data can be collected instantly.

Environmental impact

AUPA kit does not impose any environmental hazards as part of its development or working. AUPA does not utilize any hazard gases or release any gases or any other consumable components as part of its working. It does not use any chemical components; hence, it does not have any environmental impact on society or human being. The future work can be extended to machine learning approaches; such intelligent models support disease diagnosis and expert knowledge system.