Article Preview
TopIntroduction
Developing of a better alerting system for the correct diagnosis of patient’s condition plays a very imperative role in Healthcare Wireless Body Area Network (HWBAN), as false diagnosis leads incorrect decisions, which further causes problems in treatment and even became fatal for patients’ life in some cases (Ando et al., 2016; Hauskrecht et al., 2013). In life-critical HWBANs, it is always preferable to transmit accurate data within a specific time interval, as even a slight ignorance in diagnosis can make a significant difference in the decision. The effective fact with the HWBAN is its dynamic nature, where the vital signals change their mode of significance frequently. For an HWBAN system Quality of Service (QoS) means, provisioning of exact diagnosis of the health condition, because any mistake, delay or loss of critical data can become panic for the life of a patient (Kathuria, & Gambhir, 2014; Salem, et al., 2013; Chaari, & Kamoun, 2011). However, the study says that most of the alerting systems (Tóth-Laufer, & Várkonyi-Kóczy, 2014; Skubic et al., 2013) are incompetent to identify the exact condition of a patient in the specific instance of time. The main difficulties with these systems are the false alerts notification problem (Salem, Liu, Mehaoua, & Bautaba, 2014). Received data may be imprecise due to various reasons like a sensor or link error, limited resources, interference during transmission, and transmission error, etc. All of these issues may lead to generate false alerts (Haque, et al., 2015; Dickson, & Thomas, 2015). Other barriers to these kinds of systems carrying assorted traffic and limited resources (Gambhir, & Kathuria, 2018) which includes other kinds of obstacles to it. From the above-mentioned analysis it is evident that the HWBAN has a great need of an enhanced model with new thinking so that early and accurate detection of the actual condition can be done within the precise time interval.
Recently, various machine learning methods (Arumugam, & Jose, 2018, Este, et al., 2009) have fascinated the attention of many researchers in different application areas, and according to them, machine learning approaches can be used for intelligent classification purpose also. On the other hand, according to some other researchers (Dickson, & Thomas, 2015), nature-inspired algorithms are preferable in many areas due to their robust and bendy nature towards the enhancement of the various unsolved solutions. Concerning these benefits, nowadays different application areas are considering the combined methodology (Eswaramoorthy, et al., 2016) of these two mechanisms for their profit. The understanding as mentioned above encourages us to embed these two mechanisms in our system to make it acquainted with the vital atmosphere of the HWBAN. Therefore, a hybridization model is introduced here which carrying together the benefits of both Lion Hunting (LH) and Support Vector Machine (SVM) techniques and is named as LH-SVM based alerting system (LHSVMAS). Both LH and SVM have tried to make this idea successful by overcoming the remained limitations. The main objective of the LH-SVMAS is to lessen the false alert rate and provide the accurate classification of the critical packets so that they can be transmitted as early as possible with no loss towards the concerned caregiver. Unlike other protocols, it does not require user intervention hence considered as much more effective than others and the simulation outcomes proof the same in a better way. Here, the simulation was conducted in the network simulator NS-2.35.