Indian Journal of Medical Informatics. 2010; 5(1): 2

http://ijmi.org

Original Research Article

Software Development for Alert Generation in Real Time Fighter Plane Pilot Monitoring

Goutam Chatterjee¹ and Ajay Somkuwar²

Department of Electronics & Communication,
MANIT, Bhopal,
India
Email: ¹goutam.chatterjee@gmail.com, ²asomkuwar@gmail.com

Abstract:

Due to mission oriented high-G maneuvers, physiological parameters of the Fighter plane pilot undergo dramatic changes leading to stress accumulation. This work related stress in tern forces pilot to make inadvertent mistakes which often result in drastic consequence of aircraft accidents. In spite of high-G training pilots on Human centrifuge and technological support, aircraft accidents attributable to high-G related stress, regularly take place worldwide. Fighter aircraft flying missions are intensely monitored through ground controllers on voice channels, radar coverage and metreological forecasts. Yet these measures have not prevented the accidents of fighter planes which were results of human error (pilot). Human errors are often due to the reduced cognitive responses of pilots who have just undergone the high-G induced stress.

Real time stress monitoring of pilots through non-invasive wireless sensors in Body Area Network (BAN) can dramatically alter the situation where ground based servers can be alerted indicating that pilot's physiological condition is beyond normal. Preventive means then can be employed to recall pilot or abort mission. Successful implementation of BAN with the help of wireless sensors have been carried out to monitor ambulatory patients as well as life assisted products however, to implement the same on a young, healthy individual like pilot imposes a challenging task.

The solution in terms of single biomedical parameter signal from Electrocardiogram (ECG) monitoring with only essential details pertaining to high-G related stress, its synthesis, pick up technique from the pilot body for transmission over a digital data channel from BAN to ground based server; has been already demonstrated by the authors in their previous published work.

The current paper offers the research results of developing software for the Alert generation on detection of abnormal ECG signal due to +G_z acceleration induced stress in fighter plane pilots.

Keywords:

`G'; A-LOC; G-LOC; BAN; ECG; Fighter Pilot.

I. Introduction

The Fighter aircraft initially appeared in World War-I. They were developed to counter the emergence of the reconnaissance aircraft, which, as early as 1914, had proven so valuable to ground commanders that a means had to be developed to deny one's airspace to the prying eyes of enemy airmen. Fighters are aircraft intended to win air superiority by destroying enemy aircraft. The current aircraft technologies permit the application of high and sustained forces in order to achieve the desired rapid changes in velocity and direction to enable a high performance fighter plane to engage in successful air combat. This exposes the pilot to accelerative forces (both in positive & negative) for considerable time while executing his air combat task.

The definition of Newton's second law implies that accelerative force can be measured in multiples of the standard gravitational acceleration `g'. Therefore 2`g' means acceleration forces of 9.81x2 = 19.62 ms² and 3`g' means acceleration forces of 9.81x3 = 29.43 ms<sup>2</sup>. These forces are experienced by the pilot of a maneuvering aircraft and expressed in units of `g'. Velocity, acceleration and resulting centrifugal forces on human body are vectors having direction and magnitude. Both are important parameters when considering the symmetries and non-isotropic behavior of human tissues. During aerial combat, pilot can experience vertical acceleration (±G_z). With the onset of rapid vertical acceleration force (+Gz), pilot is exposed to cellular hypoxia, which produces rapid loss of brain function. Because oxygen is transported to the brain through the cardiovascular / respiratory system, an interruption in arterial blood flow to the brain leads to cerebral hypoxia. However, loss of function does not occur immediately, when blood flow ceases. There is reserve time of approximately 4 to 6 seconds before loss of brain function begins [1,2]. Physiological control of Blood Pressure (input) is based on the closed loop baroreceptor reflex which controls blood pressure through activation of the autonomic nervous system. Although very effective in compensating for upper body hypotension, the baroreceptor reflex takes time, on the order of 6 to 9 seconds, with heart level blood pressure restored in 10 to 15 seconds [3]. This compensatory response is therefore slower than the cerebral hypoxia reserve time of 4 to 6 seconds. If sufficient +Gz is experienced, the sympathetic response is inadequate and cerebral hypoxia occurs. A measure of autonomic nervous system response to +Gz to heart rate, which increases directly with increased +Gz level, reaching a maximum within a few seconds of exposure. High sustained +Gz exposures usually result in a maximum heart rate of approximately 170 beats/ min. In contrast, the parasympathetic nervous system attempts to lower upper body blood pressure by decreasing heart rate, stroke volume and total peripheral resistance. This general relaxing of myocardial and vascular tissues occurs quickly, in comparison to the sympathetic nervous system response and can be fully developed within 2 to 4 seconds [4]. During -Gz, heart rates fall dramatically; reductions up-to 50 beats/min have been recorded during exposures of -3Gz with some pilots experiencing brief period of asystote [5].

In addition to the baroreceptor response, sympathetic nervous system dominance is felicitated by the endocrine system. Physiological responses to air combat, aerobatics, centrifuge experiments or any unusual G-exposure elicit an immediate "fight or flight" response with increased levels of epinephrine, norepinephrine and serum cortisol [6]. The endocrine response is slower than the baroreceptor reflex, but becomes important as G exposure increase in duration. The respiratory system is also affected by increased +Gz. Grey-out or loss of peripheral vision occurs when systolic arterial pressure at eyelevel falls to some 50 mmHg and progresses to complete failure of vision function, or blackout, as the pressure falls to around 20 mmHg. The increased weight of the soft tissues and limbs produces significant decrements in mobility even at low levels of +Gz acceleration. It becomes impossible to rise from the sitting position at about +3Gz. With increasing +Gz, symptoms of early cognitive impairment can develop. This syndrome, termed "Almost Loss of Consciousness" (A-LOC), consists of a transient incapacitation without complete loss of consciousness. A-LOC is characterized by a blank facial expression, twitching, hearing loss, transient paralysis, amnesia, poor word formation and disorientation [7]. The most prevalent symptom is reported to be a disconnection between cognition and the ability to act.

If cerebral hypotension progresses beyond the symptoms of visual impairment and A-LOC, G-LOC can occur. It has been defined as a "state of altered perception where in (one's) awareness of sudden, critical reduction of cerebral blood circulation caused by increased G-force". During centrifuge training of pilots, those who experience G-LOC incapacitation (alter reduction of +Gz) has been divided into two periods viz. absolute incapacitation and relative incapacitation. According to centrifuge training studies [8], the average absolute incapacitation period lasts 12 seconds (range of 2 to 38 seconds). This if followed by a period of relative incapacitation consisting of confusion/disorientation that lasts an average of 12 seconds (range of 2 to 97 seconds). A pilot is unable to maintain aircraft control during either of these period averaging 24 seconds (range of 9 to 110 seconds).

II. ECG Signal Analysis for determination of G-LOC condition

All the physiological parameters exhibit linear variation with the high acceleration onset but our research of 128 case studies relating high acceleration effects of G-LOC with physiological parameters, has conclusively established that Heart rate Variability (HRV) is the only parameter which can be convincingly captured, treated, transformed and recorded through ECG monitoring for real time pilot monitoring [9]. It is still physically and technically challenging to record retinal, musculo-skeletal and respirational parameters for linking them to G-LOC symptoms. Secondly, retinal and musculo-skeletal effects even at their poorest form, do not restrict or degrade pilot cognitive responses so much that an accident may take place. For example, even when the complete black out of vision occurs, pilot may remain conscious to maintain flight profile till the return of vision. Similarly, the increased weight of the soft tissues and limbs produces significant decrements in mobility even at lower levels of +G_z acceleration. The raising of head, getting up from the seat and limb movements progressively becomes difficult but the control on fingers and palm for aircraft control switches & levers can be maintained even up to +8 G_z provided that consciousness is maintained. In contrast, cardio vascular degradation in terms of Bradycardia (slow heart) and Tachycardia (fast heart) can pronounce the impending arrival of G-LOC which is so harmful and can lead to fatal consequences.

III. Integrated Pilot Monitoring System

As is evident, a range of physiological parameters can give away the pilot condition under exposure to high Gz. There are some symptoms like heart rate variability which can single handedly forecast the impending consciousness condition and there are set of symptoms which together would determine the LOC condition. Currently, there are many integrated patient monitoring systems, which are wearable, are available through many vendors [10]. These systems have been mainly developed for ambulatory patient monitoring and life assisted products.

Commercially available integrated human monitoring systems have to be tailor made into specific applications. Choice of processor to control data rate, throughput, resolution etc. would decide the size of sensor node along with its interface for connection to a wireless medium for onward propagation to a diagnostic server [11]. The placement and number of sensor nodes decide the wireless network protocol to be chosen as output power of transmitter, distance covered for making connectivity to a neighboring sensor node, use of antenna and radiation effects are some key parameter which needs to be considered for a specific application. While Wireless Sensor Network (WSN) is a generic term used for all types of sensors, connected through a common wireless network; wireless sensors in Body Area Network (WBAN) is a specific application where power levels of transmission needs to be controlled in order to avoid harmful exposure of radiation effects on human body [12]. The reduced power levels imply careful antenna designs and avoidance of body itself as propagation medium. It also makes connectivity of sensors, a critical parameter for consideration [13]. The above cited condition makes it possible to treat the entire Integrated telemedical monitoring system to be broken down into five distinct subcategories, These are :-

(a) Multiple sensor nodes; each capable of sampling, processing and communicating one or more vital physiological signs like heart rate, blood pressure, brain activity, visual indications, oxygen saturation and musculo-skeleton reflexes. Typically, these sensors are placed strategically on the human body as tiny patches or woven into user's clothing allowing ubiquitous health monitoring in their native environment.

(b) An intra Body Area communication network which unambiguously allows sensors inter connectivity as well as ability to connect to a Mobile Base Unit (MBU) acting as personal server. The intra BAN communication will be on a chosen network protocol which would allow reliable connection among sensor nodes for steady flow of data. According to the chosen protocol, Medium Access Control (MAC) strategy is divided for physical implementation of the network. During the current research, protocol and MAC strategy were chosen based on extensive MATLAB simulation [14,15,16].

(c) A mobile base unit (MBU) which acts as personal server performs the following tasks :-

• Initialization, configuration and synchronization of WBAN nodes.

• Control and monitor operation of WBAN nodes.

• Collection of sensor readings from physiological sensors.

• Processing and integration of data from various physiological sensors providing better insight into the user's state.

• Providing an audio and graphical user - interface that can be used to relay early warning.

• Secure communication with remote diagnostic server in the upper level.

An MBU would have dual power level operation for two different types of wireless networks. In one hand, it has to collect sensing data from intra BAN at very low levels of power (10 mw or less) and one the other hand, it has to forward these data on a traditional wireless networks like GSM, Internet, Satellite link or VHF/UHF military channels. The MBU or personal server can be implemented on a PDA, 3-G cell phone or a personal computer where front end collects data at low power levels and back end acts as a gate-way for connection to extra BAN communication network.

(d) While all possible types of wireless networks can be chosen as extra - BAN communication medium, real time fighter plane pilot monitoring precludes all but VHF/VHF military wireless network. However, this link, which is used for aero-telemetry can be a single hop or double hop system. The single hop communication requires fighter plane to be in direct communication with ground based controllers but such connection can not be maintained uninterrupted for the entire duration of the flight due to terrain obstacles as well as low level flying maneuver. Therefore, the obvious choice falls on a double hop system where first communication is made to a high altitude AWAC (Airborne Warning and Control), Refueller or Jammer class aircraft; positioned at a stand off distance from the fighter formation. The modern day fighter plane formations always comprise of combinations of Jammer, Airborne Warning & Control (AWAC) and Electronics Warfare (EW) capable aircraft which remain at a safe distance from the actual fighter planes but always in communication with them. The sensory data thus first transmitted to stand off aircraft which further relays these data to the ground based controller. In this way, a secured and uninterrupted data flow can be maintained for the entire duration of the flight.

(e) Once a reliable, secured and continuous data flow from fighter plane is established to the ground controller, the diagnostic server can receive, identify and process the data for recognition of any possible physiological anomaly of the pilot and subsequent alert generation for possible mission saving efforts by the ground based controllers. The secured (encrypted) data needs to be first decrypted and then the individual pilot needs to be identified. A specific sequence of physiological data from each sensor can be predetermined at MBU as well as diagnosable server so that the flow of data is recognizable in terms of ECG, EEG, BP or EMG data. The real-time data is compared with individual pilot's restive normal physiological data which is pre stored in the server data base. Since, the tolerance levels of each individual vary; individual upper and lower thresholds for each medical parameter are defined in the server for alert generation. As and when, any of these thresholds are crossed by the real time incoming pilot data, an alarm is generated. This alarm is audio as well as visual. A time stamping of incoming data by the server ensures that in future every single bit of stored data can be timed to the occurrence of alarm generation. The incoming data can be received by the server at its TCP/IP or any other chosen port. A subroutine of the diagnostic server always senses this port for incoming data

In normal case of biomedical data communication, 20 channels of EEG, 3 channels of ECG and 1 channel of heart rate can be expected. However, in case of healthy persons like pilots, only 8 channels of EEG, 1 channel of ECG and 1 channel of respiration would be sufficient to indicate peak variation or frequency variation in the respective biomedical parameter. Since, only single contentious mode broadcast channel is proposed among the wireless sensor nodes and the Mobile Base Unit(MBU), suitable tagging of incoming data, high sampling rate and multiplexing of the incoming channels have to be provisioned by the MBU. Once the sensor data have been multiplexed, a VHF dual hop digital link can carry it to the ground based server.

The server data base will store the individual restive (no-stress) parameters for individual pilots in the similar format as it is detected and transmitted by the sensor nodes. The monitor server will identify the pilot and parameter data type before comparing them with the stored data. Individual thresholds (both positive and negative) can be set for each monitored parameter for the generation of alert. The alert can be audio as well as visual for the Flight Controller at ground. On the detection of an `alert', a series of actions can be initiated to warn the subject pilot e.g. switch to `Auto Pilot' mode, climb to safe height, maintain level flight, ground initiated recovery process in case of mechanical/electrical failures, etc.

The current paper is our last publication effort towards our research endeavor which includes extensive simulation of intra-BAN parameters on MATLAB platform, result analysis of simulation results with the help of mathematical regression tool ANOVA (Analysis of Variance) and Diagnostic software development for pilot monitoring. The present paper dwells on the last sub category of the integrated telemedical monitoring system; that is, the development of software which can capture, identify and compare the incoming biomedical data stream and generate alert to indicate impending A-LOC or G-LOC condition of the pilot.

IV. Diagnostic Pilot Monitoring Server System

The Pilot Monitoring Server System would be a cluster of servers which would track and store data records. These servers can be hosted on a single server machine or multiple server machines depending upon the scale of deployment. The diagnostic software would utilize:-

• Database server

• Monitoring server

• Reports

• Monitoring System

The Monitoring Server and the Database Server can be further clustered into distributed servers which may be physically on separate locations.

Database server:

The Database server software that we use in our scenario is the Microsoft SQL Server 2005. The choice of SQL server was due to our familiarity and ease of coding with the platform rather than any technological reasons. For that matter a designer can also deploy an Oracle or `Microsoft Access' database, if required for any reason. The database architecture of the system would look as Figure-1.

Flow Diagram of Diagnostic Server:

The flow diagram of the diagnostic server is shown in Figure-2.

As soon as the server is started, it starts monitoring the TCP/IP port for any incoming data stream. With the preamble in the data header, it establishes a connection. Any unwanted or spurious data stream connection would be rejected and it will start listening for the arrival of valid data stream. After the connection is established, it verifies the incoming data with the help of Pilot identification label and type of parameter label. Upon identification match, corresponding pilot's normal condition data is loaded from the database.

Since, the data stream can consists of several packets from different pilots, after identifying the pilot; each pilot's data will be loaded. The monitoring software is capable of processing data from several pilots simultaneously; the only restrictive parameter would be the server processing power. The incoming stream of data is compared with the normal-time parameters of individual pilots for checking if the real time data has crossed any of the previously set thresholds. If at any given time, the threshold is crossed, an alert would be generated. The alert is audio as well as visual for the purpose of ground based controllers. The generated alert is logged in the database for record keeping and to create histogram of the individual pilot. The database can have a mixed range of biomedical parameters record for different pilots. It means that every pilot need not have similar bio-monitoring. Another aspect that was recognized while creating the medical records was, that every individual has different set of normal values of ECG, EEG, pulse rate etc and also they react differently in similar conditions. In other words, threshold levels (both negative as well as positive) for every single pilot would be different for each parameter type. This necessitates that each pilot's `restive' (no-stress) medical record must be compared with incoming real-time data for processing while real time monitoring is activated. The stated objective of the database is realized by database tables.

Database Tables:

There are four database tables involved in the monitoring of pilot parameters, the `ParameterType' table, the Parameters table, Pilot-ID table and the Alert table.

Records contained in ParameterType table are shown in the Figure-3 below:

The `ParameterType' is a seed table, meaning it doesn't participate directly in database Create/Retrieve/Update/Delete (CRUD) operations. It only has fixed set of records identified at design time and modified only when system design changes at a later point of time. In the Figure-3, three monitored parameters namely EEG, ECG and BP have been indicated. They have an identification associated to it and every other table that refers to Parameters uses these identifications.

The Parameters table is shown in the Figure-4 below with its contents. This content is modified when new pilots are added to the system. This table maintains which parameters are monitored for which pilots. For example, EEG may not be monitored for Pilot-id-1 but may be monitored for Pilot-id- 2. There is a foreign key relationship between this table and `ParameterTypes' on the column `ParameterType' ID.

The two important columns in this table are the `Upper Limit' and the `Lower Limit'. This table most importantly allows individual pilot's parameter thresholds to be controlled. By setting the Upper and Lower limits, we set the thresholds for when an alert needs to be logged for a particular pilot.

The table that stores the most important data is the Alerts table. It stores the logged alerts. Figure-5 shows a data snapshot of the Alerts table. The ID field in the first column of the table is session number of data streaming. For example, in the given figure, 15 times data stream was detected by the monitoring software for pilot number-1, which is indicated in the second column of the table. The third column of the table gives the parameter identification by a numeric number. In all rows, number five indicate that only one parameter was monitored in all 15 sessions. The value field will change from zero, only in case of threshold crossing and would reflect the normalized below-threshold value. The message field in the table will record "alert" in case of threshold crossing, otherwise it will give `NULL' message in normal cases.

To summarize, the software has been made extremely innovative and versatile to accord any type of new additions or modifications in terms of `ParameterType', threshold setting and `upper' & `Lower' limits of parameters. This has been necessitated due to the known fact that huge variations in individual physiological parameters are possible, even in same age group.

Working of Pilot Monitoring Diagnostic Server:

The developed software on C Sharp DOTNET platform for Pilot Monitoring is programmed to sense any of the parallel or serial ports of the computer; however, in the present case it was set for the TCP/IP port of the computer. The software could not have been verified in the absence of actual digital data of bio-medical parameter; therefore, a TCP data simulator was generated through the C Sharp DOTNET coding. The user application window of the TCP data Simulator is shown in Figure-6. All the monitored parameters like ECG, EEG and BP are shown as `Input Parameters'. For each parameter, a normalized scale between 0 and 1 has been used. For every individual pilot, the incoming parameter data stream has been considered as normal between the two set thresholds and hence, on the normalized scale value of parameter between < 0.8 and >0.4 will not generate any alarm. However, as the parameter value goes to 0.8 or above, the

alarm would be generated. Similarly, if the parameter value goes to 0.4 or below, the alarm would be generated. In the absence of any real data, the variation in parameter value can be created by operating the GUI tabs for each input parameter and accordingly there is an alarm generation. For the pilot monitor as well, an application window for the user has been created where a user can select the pilot number, as well as the port at which he expect data stream to be connected. Figure-7 shows the pilot monitor application window.

When the Pilot Monitoring program is started, it opens up the above window for the user. The user can select a desired port by filling up the port number. In the present case, TCP/IP port has been numbered as 3333. As the monitoring software is started, it starts listening to the port for any arrival of data. Now, as the TCP data simulator program is run, it opens up the GUI window as well as the graphical window for the selected parameters. This is shown in Figure-8.

In Figure, 0.5 values are shown in green color indicating, as the tab setting on simulator was at 0.5 for all the three parameters viz. ECG, EEG and BP. The green color on the horizontal line and 0.5 indicate normal data.

Now, if any of the tab associated with ECG, EEG or BP is changed, the graphical representation would also change accordingly. For example, if ECG tab is moved on the simulator to 0.8, graphical representation will also change. Since, 0.8 values pertain to Alarm condition, the value 0.8 will appear in red color background and start flashing. The graph will also rise proportionally to the new value. Figure-9 shows new simulator setting and Figure-10 shows corresponding graphical representation.

Similarly, when any of the two parameters cross the threshold, corresponding alarms would be generated. In Figure-11, ECG and BP have crossed the upper thresholds and their respective values are shown in red alarm area. It may be noted that, EEG is within the normal values hence, no alarm is generated and it remains in green area.

However, if all the three parameters are chosen in the simulator to cross the thresholds, all three alarms will be generated. A typical setting in the Data simulator has been done, where ECG value is 0.8, the EEG value is at the top of the scale at `1' and blood pressure (BP) has been set to 0.9; which is shown in Figure-12. As soon as these new parameters with their upper limit crossed is sensed by the pilot monitoring software, it immediately generates alarm for all the three parameters. In the present case, red alarm starts flashing for ECG, EEG and BP which is shown in Figure-13.

As explained in section-I that, heart rate and pulse rate of the pilot go down when he experiences negative acceleration (-Gz); a different color coding has been used to indicate this condition. Accordingly, if the real time parameters cross low threshold, the alarm coding has been done in Orange color. This instantaneously indicates to the ground controller the maneuver type which pilot is currently undergoing. Figure-14 indicates data simulator tab setting for lower threshold generation for all the three parameters and Figure-15 shows the corresponding alarm condition.

Therefore, TCP data simulator very importantly established that the Pilot Monitoring Software was fully functional. It can detect incoming data stream, recognize the parameter type and detect the upper as well as lower thresholds. It was also proved that it could generate separate alarms for each parameter type and their respective thresholds. It was an important step towards actual (live) data monitoring.

Real Data Acquisition & Software Validation.

The pilot monitoring software for alert generation, upon detection of abnormality in ECG as a direct consequence of onset of +G_z acceleration, was tested against the TCP Data simulator for its trial runs. However, the validation process could not have been complete unless the real ECG data were obtained from the pilots, who are actually undergoing under the high-G stress. Since, the Indian Air Force (IAF) does not have tie ups with the Universities and academicians unlike United States Air Force (USAF) and some advanced countries; allocation of aircraft, pilot and permission to test modifications were not feasible. The only available alternative was to obtain data from the Institute of Aviation Medicine (IAM), Bangalore. IAM is a premium IAF institution, dedicated to aviation medicine researches and Centrifuge training of the fighter pilots of IAF.

V. Methodology of Centrifuge Training

The Institute of Medicine conducts Centrifuge training of fighter Pilots for three purposes:-

• To increase + G_z tolerance of aircrew so as to endure the high-G induced stress in a manner to retain the ability to control the aircraft.

• To conduct research on various anomalies developed due to flying in general and due to + G_z in particular.

• To create data repository of collected medical records during the Centrifuge training.

The purpose of the data repository is to improve +Gz-training methods and establish a large sample source of aircrew G tolerance characteristics. The repository consists of three sections. These sections are:-

• Trainee description;

• +Gz profile description;

• Cardiovascular response of the trainee to that +Gz profile.

The first two sections are obtained from Run Sheet forms completed by the trainees and training personnel. The latter section is obtained from review of the trainee's electrocardiogram (ECG) which included 2 channels of ECG data based on sternal and bi-axillary lead placement. The records are compiled for later review and entry of cardiovascular data into the database. All trainees considered for centrifuge training had either passed the appropriate physical examination requirements required for flight duty (Class I or II) or were required to be fit for +Gz exposure by a qualified physician prior to insertion in the centrifuge. The training is called the "G Tolerance Improvement Program" (GTIP). The GTIP training protocol consisted of an approximately 2-hour lecture on the effects of +Gz stress on human physiology and the methods to combat this stress. Following this lecture each trainee was exposed to the acceleration environment in the human-use centrifuge for approximately 15 minutes. While trainees underwent training in the centrifuge, their classmates witnessed the sequence of events thereby allowing them to learn from others and discuss the various techniques to better withstand +Gz. The GTIP program does not include a pass/fail valuation within its training protocol. However, the competitive nature of aircrew ensures fierce competition among themselves to prove, who was able to withstand the onset of G-LOC better.

Upon termination of the +Gz exposure, each trainee is debriefed signifying the completion of the training day. Technology to help combat acceleration stress includes the Anti-G Suit (AGS). The AGS is a protective garment used by aircrew to enhance tolerance to +Gz forces experienced in high performance aircraft. The AGS is a pair of trousers composed of inflatable air bladders strategically located over the calves, thighs, and abdominal areas. The subject dons the AGS, which is connected via a hose to an anti-G valve. The suit is activated in accordance to G exposure level. Activation of the AGS (inflation of the air bladders) exerts pressure against the lower limbs and abdomen. This pressure aids the cardiovascular system to maintain adequate blood flow to the CNS by forcing blood towards the head "counteracting" the effect of +Gz [ ]. Benefits of the AGS include increased G tolerance (G level), increased G exposure duration, and reduced petechia hemorrhage incidence. The AGS also aids the wearer to perform Anti-G Straining Maneuvers (AGSM) as it provides the subject something to "strain against" and decreases the fatigue mostly generated by this effort. Resting G-tolerance levels as defined by loss of peripheral vision have been reported to average 3 G without and 4.5 G with an AGS.

The AGSM provides further protection against +Gz stress as follows. The "Level-1" AGSM combines a periodic, 3 second strain (valsalva) against a closed glottis. A rapid exhalation and inhalation of < 0.5 seconds interrupts this strain. Tensing of all major muscle groups of the abdomen, arms, and legs is part of the effort. The AGSM provides an average 1.5 G protection above resting G tolerance levels (with AGS). This maneuver is also known as the "Hook" maneuver where the subject says the word "hook" as he begins to strain thereby ensuring a completely closed glottis. The GTIP profile includes several exposures to +Gz with and without Anti-G Suit.

Method of G onset for Centrifuge Training.

A gradual onset rate (GOR) exposure is the first of a series of +Gz runs during a single GTIP training day. The GOR exposure commences at a resting level of approximately 1.0 +Gz and increased at 0.1 G/s. Visual decrement is subjectively assessed by the trainees' inability to see an array of light emitting diodes (LEDs) placed in an arc describing 15⁰ increments (150⁰ total) 30 cm in front of the subject at eye level. The run proceeds until the trainee experiences 60⁰ Peripheral Light Loss (PLLI), i.e., vision is confined to an arc describing 30⁰ either side of the central point directly in front of the trainee. Once PLLl is reached, the trainees are instructed to perform AGSM until peripheral vision is once again reduced to 60⁰ (PLL2). The trainees then terminate the +Gz exposure by pressing a pre-selected button located on the control stick. The limit of the +Gz exposures is 6.5 +Gz. The profile is "open-loop" in that the trainee is not in control of the device but could stop the run at any time. A standard AGS is worn and activated during the +Gz exposure. The inflation rate of the suit is approximately 1.5 pounds per square inch (psi) per +Gz. Inflation of the AGS bladders starts at approximately 2 +Gz to a total of 11 psi when fully inflated at 6.5 +Gz.

The GOR exposure is followed by a series of Rapid-onset-Rate (ROR) exposures. The ROR exposures commence at a resting level of approximately 1.0 +Gz. It is then increased at approximately > 6.0 G/s (Haversine profile) until reaching a predetermined plateau level ranging from 5.0 to 6.5 +Gz. The plateau was to be maintained for a period of 10, 15, or 30 s. Both the G level and the plateau duration depended on the protocol of the time. That is, the protocol for the ROR exposures changed over the five years of training exercises under discussion. The ROR exposures are performed in sequence where the G level is increased by 0.5 or 1.0 G depending on the trainee's performance as the training exposure progresses. Most trainees experience a total of four ROR exposures to -2, 4, 6, and 6.5 +Gz. The rest period between the runs is 1 minute or more. This period allows the trainees to return to a resting heart rate similar to the one immediately prior to commencing the training runs. The AGS is worn and activated during the entire ROR exposure. The trainees are instructed to perform AGSM throughout these exposures. The ROR profile is also open loop.

Reasons for GOR and ROR run termination includes 60⁰ peripheral light loss, pain, fatigue, and G-induced Loss of Consciousness (G-LOC). Otherwise, the run is considered as "completed." The centrifuge's cockpit configuration is similar to the IAF used high performance training aircraft. A qualified physician monitors all exposures. Two channels of ECG activity are recorded during all runs (sternal and bi-axillary lead placement).

VI. Sample Data Details

The ECG sample data received for a batch of 10 pilots were taken during ROR cycle of 44 seconds. The ages of pilots were between 24 -29 years and each pilot had flown at least 2000 hours on fighter aircraft. Each one of them had reported at least one A-LOC/G-LOC incidence and this was their first Centrifuge training. During this particular training day, no Anti G Suit was worn and +G_z exposure was unaided. As per the medical records, none of the pilots had ever displayed the symptoms of Bradycardia or Tachycardia. The data values were converted to R-peak records with the help of previously described wavelet transform, statistical algorithm and low pass filter. Then these R-peak values were converted to numeric values. A real time data feeder program on C-Sharp/DOTNET was written to read these numeric values line by line and convert them into graphical display. As the ECG data was taken for 15 seconds at + 1 G_z, for 15 second at - 2 G_z and 14 seconds at + 4 G_z; it was imperative that these onset values of +G_z are also displayed along the ECG graph. Therefore, acceleration rate change was also programmed to be displayed along with the ECG values. The graphical display of 44 second data pertaining to a pilot is shown in Figure-16.

In the figure, R-peaks are shown in the ECG window. The entire window is of 44 second. During this period, ROR +G_z ensure that acceleration rate changes three times. The disposition of +G_z state is shown in the window marked acceleration. Initially when the acceleration was maintained at + 1 G_z, 17 R-peaks appear in 15 second period indicating 68 beats per minute heart rate. With - 2 G_z setting within 0.6 seconds lasts for next 15 seconds. Now the 11 R-peaks appear during this period, indicating that heart rate has slowed to 44 beats per minute. Next ROR ensures that + 4 G_z is set within 0.6 seconds but next two second periods are indeterminate in the changing state. There after heart rate settles down to 85 beats per minute as we can count 17 R-peaks during 12 second period.

When the pilot data pertaining to all the ten pilots were graphically examined with the help of Real Data feeder program, the changes in the heart rate with the onset of different +G_z were noted and tabulated which is given in the Table-1.

The following conclusions were drawn after examination of the above tabulated pilot ECG data:-

• Being young, healthy and in similar age group, heart rates of all the pilots were very close to each other.

• Their heart rate variation with the onset of +G_z follows similar pattern of going low with -2G_z and going high with + 4G_z.

• The R-peak rate variations were clearly identifiable with changing accelerations.

With the above enumerated knowledge, now it was possible to set minimum and maximum thresholds to generate alert. It was decided to normalize the heart rates on the scale of 0 - 1 where 0.1 indicated beat rate of 10 or below, 0.2 indicating 20 or less beats and likewise going up the scale 0.9 indicating beat rate of 90 to 81. Here, it is noteworthy that `normalized' does not refer to `Normal Distribution' or the terms `mean' and `standard deviation' associated with it. Considering the heart rate of 60- 80 as normal sinus rhythm, the server will not generate any alarm. Although, with heart rate going up to 90, condition of G-LOC or A-LOC is not likely to occur, such threshold was chosen to be implemented to verify the Pilot Monitoring Software's ability to generate alarm.

The chosen threshold values are in the hands of designer and it can be even chosen to personalize for individual pilots. The decision to accord flexibility in design has been deliberately taken as it is a known fact that the individual physiological parameters vary a great deal depending upon factors like region of birth place, life style and exercise regime etc. In our prototype model with the set threshold values, all the ten pilot data were fed to the monitoring software. The software was able to generate both, the low threshold alarm as well as the high threshold alarm.

The experiment was repeated many times and each time, at the correct +G_z acceleration, software was able to generate alarm. All the developed programs, including actual ECG sample data of pilots, are available with the authors for the demonstration.

VII Conclusion

This research has allowed the study of a new and very promising kind of wireless networks; the Body Area Network, for the novel application of real time pilot monitoring. Along the way, effects of high acceleration on human physiological parameters, their measurement and ECG signal pickup from the human body was studied in depth. The developed software has conclusively displayed the Alert generation capability with changing biomedical parameter under stress. It has successfully identified the heart rate variability when +G_z acceleration induced stress was caused.

In conclusion, it may be said that the real time fighter plane pilot monitoring may resolve a long pending issue of pilot safety during high-G acceleration. It would also create automation of Flight Safety in Indian Air Force, which otherwise spends hundreds of crores of rupees to ensure accident free flying. The entire concept was vigorously analyzed, simulated and tested on a software platform and now is ready to be implemented as funded project.

References

1. Whinnery JE and Shaffstall RM. Loss of consciousness during air combat maneuvering. Journal of Aerospace Safety 1978; 34: 23-25.

2. Whinnenry JE. Theoretical Analysis of Acceleration induced central nervous system ischemia. IEEE magazine of Engineering in Medicine and Biology 1991; 10 : 41- 45.

3. Bean JW. Effects of oxygen at increased pressure. Physiological Reviews 1945; 25 : 1-147.

4. Glaister DH et al. Emotional and cardiovascular stresses of centrifugation: effect of beta receptor blockade on heart rate response. AGARD conference on influence of contrasting stress as performance, 14-17 May 1972, Washington, D.C., proceeding report no.101: 76-85.

5. Harrison MH and Gibson TM. British aviation medicine during the Second World War, part-2, RAF-IAM report 1981; 610: 18-19.

6. Nash PR et al. G-induced loss of consciousness in high performance fighter aircraft. Presented at the annual scientific meeting, Aerospace medical Association, Washington D.C., 14 May1979; report no. 709-122-4415: 156-159.

7. McGown DG. A-LOC- Almost Loss of Consciousness and it's importance to Fighter Aviation. Aviat Space Environ Med 1997; 68 : 24-28.

8. Burton RR and Whinnery JE. Operational G-induced loss of consciousness: Something old; something new. Aviat Space Environ Med 1985; 58: 761-767.

9. Dehart R.L, Lea and Febiger (editors). Fundamental of Aerospace Medicine Chapter-4 (Physiology and Environment). Philadelphia: Lippincott Williams & Wilkins, 1985. pp 90-147.

10. Mitchael K et al. Application of Human trans power implants. Mobile Eighth world multi-conference on semantic, Cybernetics and information, Orlando, Florida, 18-21 Jul 2004. 505-512.

11. Chatterjee G and Somkuwar A. Design Analysis of wireless sensors in BAN: Simulation Version-1. MIC-MCA, IEEE International conf, 08-10 Aug 2008, University of Amman, Jordan IEEE Xplore pub date Aug08. 71-77.

12. Konstantas D et al. Mobile Patient Monitoring: The Mobihealth system. International Congress on Medical and Care Compunetics, NCC, The Hague, 2-4 June 2004.

13. Karnik A and Kumar A. Distributed Optimal self organization in ad hoc wireless sensor network. IEEE/ACM transactions on networks(TON) 2007. 5:5: 1035-1045.

14. Chatterjee G and Somkuwar A. Analytical Results of Intra BAN parameters through Linear Modeling. IJ-ETA-ETS, Jul09-Dec09, Amoghsiddhi publications, Sangli, Maharashtra, 2 : 386-391.

15. Lee A and Mason M. MATLAB Simulation for Computing Probability of Detection. Proceedings of IEEE Radar Conference, Pamona, California, USA, 24-26 Mar2002. 478-483.

16. Lamprinos IE et al. A low power Medium Access Control protocol for wireless medical sensor networks. IEEE Bio Medical publications 2004. 1: 23-26.

Paper received on 15/12/2009; accepted on 05/01/2010

Correspondence:

Goutam chatterjee

Department of Electronics & Communication,

MANIT, Bhopal,

India

Email: goutam.chatterjee@gmail.com

This Open Access article is available at: http://ijmi.org/index.php/ijmi/article/view/y10i1a2

© 2010 Author(s); licensee Indian Journal of Medical Informatics under

Creative Commons Attribution-No Derivative Works 3.0 License .