CC BY
80
A NOVEL CLUSTERING METHOD FOR WIRELESS BODY AREA SENSOR NETWORKS USING FUZZY LOGIC
Yahya Al-Naggar,
Ph.D. student at department of communications networks and data transmission, the Bonch-Bruevich Saint-Petersburg State University of Telecommunications, Russia, Saint-Petersburg, Russia, [email protected]
Keywords: Wireless Body Area Sensor Networks, Residual energy, Network lifetime, Stability period of network, Throughput of network, Clustering, Cluster Head Selection, Fuzzy Logic.
In this paper, we propose a new method that combines cluster formation and cluster head selection in wireless body area sensor networks (WBASN). Applications of wireless body sensor networks include health monitoring as well as control of the medical parameters in the sports industry. Since network topology in the standard IEEE 802.15.6 is defined as one hop star topology, this paper focused on one hop star topology and a multiple hops cluster-based topology for WBASN. A cluster-based topology WBASN meaning two hops. In а cluster-based topology WBASN, a cluster header (CH) forwards the received data packet from sensor nodes to the coordinator. We have used fuzzy logic system based on two parameters, so that we can select cluster heads. These two parameters are remaining energy and distance to the coordinator. Residual energy parameter balances the energy consumption among the sensor nodes while distance parameter ensures successful packet delivery to the coordinator. Sensors are heterogeneous, and all these sensors integrate into the human body. The number and the type of sensors vary from one patient to another depending on the state of the patient. The most common types of sensors are the "Electroencephalography (EEG)" which is used to measure the electrical activity produced by the brain, the "Electrocardiogram (ECG)" which is used to record the electrical activity of the heart over time, the "Electromyography (EMG)" which is used to evaluate physiologic properties of muscles, Blood pressure, heart rate, glucose monitor, the Pulse Oximetry (SpO2) which is used to measure the level of oxygen saturation in the blood, to measure temperature of the body, respiration and motion (Gyroscope/Accelerometer), etc. Our simulation results show that the proposed method consumes less energy, thereby prolonging the network lifetime, and enhances the stability period of network and packet delivered to the coordinator compared to star topology method.
Для цитирования:
Аль-Наггар Яхья. Новый способ кластеризации с использованием нечеткой логики для нательных беспроводных сенсорных сетей // T-Comm: Телекоммуникации и транспорт. - 2015. - Том 9. - №6. - С. 78-83.
For citation:
Yahya Al-Naggar. A novel clustering method for wireless body area sensor networks using fuzzy logic. T-Comm. 2015. Vol 9. No.6, рр. 78-83. (in Russian).
W
T-Comm ^м 9. #6-2015
Introduction
In general, wireless body area sensor networks (WBASNs) are wireless networks that support the use of biomedical sensors and are characterized by: very tow transmitting power to coexist with other medicai equipments and provide efficient energy consumption; high data rate so that they can allow applications with high QoS (Quality-of-Service) constraints; low cost, low complexity and miniature size to allow real feasibility [1].
Sensors are heterogeneous, and all these sensors integrate into the human body. The number and the type of sensors vary from one patient to another depending on the state of the patient. The most common types of sensors are the "Electroencephalography (EEG)" which is used to measure the electrical activity produced by the brain, the "Electrocardiogram (ECG)" which is used to record the electrical activity of the heart over time, the "Electromyography (EMG)" which is used to evaluate physiologic properties of muscles, Blood pressure, heart rate, glucose monitor, the Pulse Oximetry (Sp02) which is used to measure the level of oxygen saturation in the blood, to measure temperature of the body, respiration and motion (Gyro-scope/Accelerometer), etc. [1, 2, 5].
In such networks different kinds of sensors are attached on clothing or on the body or even implanted under the skin. For example, patients do not need be physically present at the physician clinic for their routine diagnostic check if they are equipped with WBASN.
The main aim of this paper is to evaluate/analyse the four different types of experiments based on WBASNs stated below:
• Enhancing network lifetime
• Enhancing stability period of the network
• Minimizing energy consumption of the nodes
• maximizing throughput of the network
Wireless body area sensor network architecture
A WBASN architecture shown in Fig. 1 consists of sensor nodes, a coordinator and communication channels for transmitting gathered signal information over a wireiess network to the control center [5, 6].
The classification of nodes in WBASNs based on their role in the network is as follows [2, 5]:
Coordinator - The coordinator node is like a gateway to the outside world, another WBASN, a trust center or an access coordinator. The coordinator of a WBASN is the personal digital assistant (PDA), through which all other nodes communicate.
Sensor - Sensors in WBASNs measure certain parameters in one's body either internally or externally. These nodes gather and respond to data on a physical stimulus, process necessary data. Some existing types of these sensors could be used in one's wrist watch, smart mobile phone and consequently, allow wireless monitoring of a person anywhere, anytime and with anybody. The wireless sensor nodes collect information of a body, relay it through the coordinator and store the biological information on the data center through the communication infrastructure.
Emtrgtnn f Ambulance
Fig.l. Wireless body area sensor network architecture
Wireless body area sensor network system
One of the monitoring systems is wireless body area sensor network. A WBASN system can be divided into 2 schemes; scheme 1: all sensors transmit the signal directly to the coordinator via one hop, scheme 2: the sensor transmits the signal to the coordinator via multiple hops [3].
Star Topology
The sensor should use high power to transmit the signal because the coordinator isn't always close to sensor. Therefore, the life time of sensors becomes shorter and each sensor causes the interference to other sensors in its area. Moreover, the connection between sensors and the coordinator may fail due to the interruption of the body, especially when the human is moving. All sensors transmit the vital data toward the coordinator. In WBASN system based on one-hop star topology, all sensors transmit its data directly to the coordinator. The vital data packet is generated at each sensor by its access probability.
Cluster-Based Topology
Clustering network is an efficient and scalable way to organize WBASNs, Appropriate cluster-head selection can significantly reduce energy consumption and prolong the lifetime of WBASNs. In clustering, the sensor nodes are divided into some clusters and one node is selected as duster head in each cluster. Since each sensor transmits the signal to the neighboring sensor, the transmit power, the transmit area and the effective area are small. Therefore, the number of interfered sensors decreases and the lifetime of sensors increases. In additional, even if the direct connection between the sensors and the coordinator fails, the sensor can transmit to the coordinator via other sensors that connects to the coordinator. According to the definition of network cluster-based topology two hops is considered. Member sensors transmit the data packets to their CH and CHs forward the received data packets to the coordinator. However, in a cluster-based topology, a sensor can transmit the signal to its cluster header (CH) instead of the coordinator. Since all sensors generate a
packet of vital data and transmit forward to the coordinator via the CH, the transmit power of all sensors and the throughput is expected to be improved.
In this paper, we assume S sensors are deployed on the human body. We represent the i-th sensor by S| and consequent sensor node set s = si,s2,.....,sn.
■ We deploy sensor nodes on the human body at fixed places.
• We place the coordinator at waist.
■ Sensor nodes and the coordinator are stationary after deployment.
■ We use heterogeneous sensor nodes.
■ All sensor nodes have the same initial energy.
■ We use first order radio model proposed in [4]. This radio model considers d as the separation between transmitter and receiver and d2 as the loss of energy due to transmission channel. First order radio model equations are given as:
ETx(k,d) =
^Tx-elec(^) "I" ^Tx-amp
ETx(k, d) = ETx_eiec x k + Eamp x k x d2 Enx(k) = ERx_e,ec(k) x ERx(k) ErxCO = ERx_e|ec x k
(i) (2)
where ETx is the energy consumed in transmission, ^Rx is the energy consumed by receiver, ETx-elec and ^rx-else are the energies required to run the electronic
r
circuit of transmitter and receiver, respectively. tfl»V is
the energy required for amplifier circuit, while ^ is the packet size.
Clusters formation
In our method, the formation of clusters is as follows sensing field (human body) is divided into two logical areas: first area of 0-100 cm and the second region of 100200 cm. In each round, selection of the cluster head is based on fuzzy logic. After the cluster head selection, the distance between the sensor node and the cluster head in each cluster is computed as shown in equation (3).
.2
distance (SifCj) = qf
(3)
where is the sensor node in cluster (i=l,..,,m) and ci is the cluster head (j-l,...,k).
The distance between the cluster head and the coordinator is calculated using the equation (4).
distance (Coordinator, CH ) = J(x CH xcoor) + (yCH — Ycoor)
(4)
where *ch and xC00r are the points of cluster head and the coordinator on the Axis, YcH an(i Ycoor the points of cluster head and the coordinator the on Axis.
Selection of cluster heads using fuzzy Logic system
The concept of fuzzy set and fuzzy logic was introduced by Zadeh in 1965 [8]. The structure of a fuzzy logic system is shown in Fig.2. When an input is applied to a fuzzy logic system, the inference engine computes the output set corresponding to each rule. The defuzzifier then computes a crisp output from these rule output sets.
Rules
Crisp inputs
Fuzzifier
Fuzzy input sets
Defuzzifier
Inference
Fuzzy output sets
Crisp Output
Fig.2. The structure of a fuzzy logic system
We use Mamdani-style fuzzy inference in this paper. The process is performed in four steps:
> Fuzzification of the input variables. Take the above input variables and determine the degree to which these inputs belong to each of the appropriate fuzzy sets.
> Rule evaluation. Take the fuzzified inputs, and apply them to the antecedents of the fuzzy rules. Because the given fuzzy rule has multiple antecedents, the fuzzy operator (AND) is used to obtain a single number that represents the result of the antecedent evaluation.
> Aggregation of the rule outputs. Take the membership functions of all rule consequent previously clipped or scaled and combine them into a single fuzzy set.
> Defuzzification. Evaluate the rules, but the final output of a fuzzy system has to be a crisp number. We use the center of gravity (COG) as the defuzzification method and it is expressed as shown in equation (5):
Output (PCHS) =
J * Ha 00 dx / ha (x)dx
(5)
where ^a fx) is membership function of set A.
Once clusters formation is complete, we proceed to the selection of cluster heads using a system based on fuzzy logic [8]. Table 1 consists of Inputs and output variables with their associated Fuzzy values.
For the primary fuzzy system of this paper, two Input linguistic variables are defined, representing the remaining energy and distance to the coordinator. For remaining energy and distance to the coordinator, the term set is defined by three labels, so the fuzzy inference rule has 3x3 = 9 rules. Moreover, the system has one output linguistic variable named Probability of Cluster Head Selection (PCHS). The term set of output linguistic variable is divided into nine levels. The fuzzy if-then rules in the Cluster Head Selection are also shown in Table 2.
T-Comm Tом 9. #6-2015
Table!
Inputs and output variables with their associated Fuzzy values
System's Linguistic Linguistic Fuzzy
Variables variables values intervals
Remaining Low 0-0.25
Energy Medium 0.15-0.35
[0 - 0.5] J High 0.25-0.5
Inputs Distance to Close 0-60
the coordi- Middle 40-80
nator Far 70-100
fO - 1001 cm
Very small 0-15
Small 5-25
Probability Rather 15-35
Y of small 25-45
Cluster Head Med. small 35-65
Output Selection Medium 55- 75
[0 - 100] % Med. large 65-85
Rather 75-95
large 85-100
Large
Very large
Table2
The fuzzy if-then rules in the Cluster Head Selection
Rule IF AND THEN
Num- Remaining Distance to Probability of
ber Energy the coordina- Cluster Head
tor Selection
1 Low Far Very small
2 Low Middle Smail
3 Low Close Rather small
4 Medium Far Medium small
5 Medium Middle Medium
6 Medium Close Medium large
7 High Far Rather iarge
8 High Middle Large
9 High Close Very large
For example, we can read a rule in the following manner: if (the remaining energy is high) and (distance to the coordinator is close) then (the Probability of Cluster Head Selection is very large).
The rules are created using the Fuzzy Inference System (FIS) editor contained in the MATLAB Fuzzy Toolbox [7, 8]. Fig.3 shows a sample fuzzy calculation of a probability of cluster head selection based on the amount of remaining energy and distance to the coordinator. Fig.4 shows a control surface.
Rfioaioiug energy = 0.4 J
Distance lo Coordinator = 45 cm
Probability of Cluster Head Selection = 91.5 %
Fig.3. Sample fuzzy calculation of a probability of cluster head selection
Remaining energy , J
Distance to coordinator, cm
Fig.4. Surface view of probability of cluster head selection with respect to remaining energy and distance to the coordinator
Remaining energy, J
5 J
9
3
I :
a ^ 1
Distance to coordinator, cm
Fig.5. The probability of cluster head selection
Fig. 5a shows simulation results with 40, 70 and 90 values for the distance to the coordinator where it is evaluated for probability of cluster head selection and remaining energy. And Fig. 5b shows the simulation results considering remaining energy values of 0.10, 0.25, and 0.40, where the probability of cluster head selection and distance to the coordinator is evaluated.
Simulation results and analysis
In this section, we present the simulation results were carried out in MATLftB. The simulation parameters with their values are shown in table 3.
Sensor iodes deployment ou the human body
Table3
Simulation parameters with their values
Type Parameter Value
Network Number of sensor nodes (n) 10
topology Probability of cluster head (P0Dt) 0.2
Network coverage (0, 0) (100, 200)cm
The height of human body 180 cm
The width of human body 60 cm
Coordinator location (45, 110) cm
Radio Initial energy Eq 0.5 Joules
model Energy for data aggregation EDA 5 nJ/bit/signal
Transmitting energy (Ete-eiec) 16.7 nj/bit
Receiving energy (E№eiec) 36.1 nJ/bit
Amplification energy (EamD) 1.97 nJ/bit/m2
Data packet size (k) 4000 bits
In the network model we used a model of 10 nodes which is distributed at fixed locations on the human body 60 * 180 cm, as shown in Figure 6 (a). The simulation result with a direct connection (star typology) and the simulation result via cluster-head connection (cluster-based topology) are shown in Figure 6 (b, c).
In order to evaluate our method, we used the stability period of the network, network lifetime, throughput and residual energy and these are defined in detail here:
Network lifetime: It represents the total network operation time till the last node dies.
Stability period of network: Stability period is the time span of network operation till the first node dies. The time period after the death of first node is termed as unstable period.
Residual Energy: In order to investigate the energy consumption of nodes per round, we consider residual energy parameter to analyze energy consumption of network.
Throughput of network: Throughput is the total number of packets successfully received at coordinator.
E 140 -Ï-
2 12a
5
i
4
1 m i
»
ï
5
1 99
. HO
£ 120
^ 100
i
$ 80
O Seusor nodr
J SA
□
o
/ Vi j
The vridth of human bodv, cm
-a)
Sinilitioi muh t ta dirKt e {Star Topology)
O Su □ t- nor nod* ötdüittm
\ t 1 ):
O .....« ! i ! ■ I
1* °
r / i jj \ i i À 1
Ii V. i! ! \>J
/ / U »V
i
20 40 60 00 100
The width of fa um jo bod\. cm
Simulation result via cluster-head connection (Cluster-Based Topology )
b)
o o Sensor node
■ Coordlnitor V )i
« ;
f
J I A' \ T
' V« I
The width of human !>odvr cm
Fig.6. The simulation results tri MATLAB Figures 7, 8 and 9 show the simulation results for cluster-based topology method in comparison with star topology method.
Stability period of network & Network lifetime
Chn№ based topologv method Slar lajiology ir. ft Noil
Star I ; I m >11 >1: \ m t'F h mt
1 i il '-U' I hnsed (npotngy ni I't tiod
3000 4000 5000 Number of rounds (r)
Fig,7. Number of alive sensor nodes vs. number of rounds Residual Energy of network
5
4.5 . 4
¡3.5 3
B 2.5 2
1.6 1
0.5
fr-bawd topology met opology method
- Clusl iod
Start
3000 4000 6000 Number of rounds (r)
<1.5 4
= 3.5 ^ >-
r> £ S s 3
n S2.S S F
Fig.8. Remaining energy vs. number of rounds 10J Throughput of network
II
: • • ■
- Star topology method
a 1000 2000 3000 4000 5030 6000 7000 8000
Nmnlwr of rounds (r)
Fig.9. Number of packets successfully received at coordinator vs. number of rounds
Fig. 10 shows the number of rounds until the death of the first sensor node and the last sensor node for cluster-based topology method in comparison with star topology method, Sensor node is considered dead, having exhausted its energy.
.Stability period of network Network lifetime
Fig. 10. Node death comparison methods
Conclusion
In this paper we have proposed a new method that combines cluster formation and cluster head selection in wireless body area sensor networks. We have used fuzzy logic system based on two parameters: remaining energy and the distance to the coordinator to select cluster heads. The sensor nodes collect data and send it to their duster heads then from cluster heads to the coordinator. Simulation results showed that, the proposed method increases the stability period of network approximately 3 times and extends network lifetime approximately 2 times compared to star topology method. Our model and results show that the presented method performs well with respect to the length of lifetime and the stability of the network, and also with respect to residual energy and throughput of the network.
References
1. Chehri, A., Hussein T. Moutah. Survivable and Scalable Wireless Solution for E-health and Emergency Applications. // In EICS4MED 2011. Proceedings of the 1st International Workshop on Engineering Interactive Computing Systems for Medicine and Health Care. Pisa, Italy. 2011. pp.25-29.
2. Garth V. Crosby, Tirthankar Ghosh, Renita Murimi, Craig A. Chin. Wireless Body Area Networks for Healthcare: A Survey // International Journal of Ad hoc, Sensor & Ubiquitous Computing (IJASUC) Vol.3, No.3, June 2012, pp. 1-26.
3. Pham Thanh Hiep, Nguyen Huy Hoang, Maximizing Throughput of Cluster-Based WBAN with IEEE 802.15.6CSMA/CA // Internationa! Journal of Multimedia and Ubiquitous Engineering Vol.9, No.5 (2014), pp. 391-402.
4. Q, Nadeem, N. Javaid, S. N. Mohammad, M. Y. Khan, S. Sarfraz, M. Gull. SIMPLE: Stable Increased-Throughput Multi-hop Protocol for Link Efficiency in Wireless Body Area Networks. // Broadband and Wireless Computing, Communication and Applications (BWCCA), 2013 Eighth International Conference on, vol., no., 28-30 Oct. 2013, pp.221-226.
5. Movassaghi, S.; Aboihasan, M.; Lipman, J.; Smith, D.; Jamaiipour, A. Wireless Body Area Networks: A Survey. // Communications Surveys & Tutorials, IEEE (Volume: 16, Issue: 3), 14 Jan. 2014, pp. 1658-1686.
6. Shah Murtaza Rashid A! Masud. Study and Analysis of Scientific Scopes, Issues and Challenges towards Developing a Righteous Wireless Body Area Network // International Journal of Soft Computing and Engineering (IJSCE) ISSN: 2231-2307, Volume-3, Issue-2, May 2013, pp. 243-251.
7. MathWorks - MATLAB and Simulink for Technical Computing, Fuzzy Logic Toolbox, Documentation Center, http://www.mathworks.com/help/ fuzzy/index, html,
8. Fuzzy Logic Toolbox'" User's Guide © COPYRIGHT 1995-2012 The MathWorks, Inc. 351 p.
