An Efficient Backoff Algorithm for IEEE 802.15. 4 Wireless Sensor ...

Wireless Pers Commun DOI 10.1007/s11277-013-1454-8

An Efficient Backoff Algorithm for IEEE 802.15.4 Wireless Sensor Networks Zahraa Dahham · Aduwati Sali · Borhanuddin M. Ali

© Springer Science+Business Media New York 2013

Abstract IEEE 802.15.4 is one of the most prominent MAC protocol standard designed to achieve low-power, low-cost, and low-rate wireless personal area networks. The contention access period of IEEE 802.15.4 employs carrier sense multiple access with collision avoidance (CSMA/CA) algorithm. A long random backoff time causes longer average delay, while a small one gives a high collision rate. In this paper, we propose an efficient backoff algorithm, called EBA-15.4MAC that enhances the performance of slotted CSMA/CA algorithm. EBA15.4MAC is designed based on two new techniques; firstly, it updates the contention window size based on the probability of collision parameter. Secondly, EBA-15.4MAC resolves the problem of access collision via the deployment of a novel Temporary Backoff (TB) and Next Temporary Backoff (NTB). In this case, the nodes not choose backoff exponent randomly as mentioned in the standard but they select TB and NTB values which can be 10–50 % of the actual backoff delay selected by the node randomly. By using these two new methods, EBA-15.4MAC minimizes the level of collision since the probability of two nodes selecting the same backoff period will be low. To evaluate the performance of EBA-15.4MAC mechanism, the network simulator has been conducted. Simulation results demonstrate that the proposed scheme significantly improves the throughput, delivery ratio, power consumption and average delay. Keywords

WSNs · MAC · IEEE 802.15.4 · CSMA/CA · Efficient backoff algorithm

Z. Dahham (B) · A. Sali · B. M. Ali Department of Computer and Communication Systems Engineering, Faculty of Engineering, University Putra Malaysia, 43400 UPM, Serdang, Selangor Darul Ehsan, Malaysia e-mail: [email protected] A. Sali e-mail: [email protected] B. M. Ali e-mail: [email protected]

123

Z. Dahham et al.

1 Introduction In recent years, the demand for Wireless Sensor Networks (WSNs) has increased tremendously and gained world-wide interest. Medium Access Control (MAC) protocol plays a vital role in the performance of WSNs. IEEE 802.15.4 MAC protocol is designed to achieve the characteristics of low power and low rate wireless personal area networks (LR-WPANs) [1]. Furthermore, IEEE 802.15.4 defines the specifications of both PHY and MAC sub-layer to meet the requirements of sensor networks. It can operate with two different channel access methods; a beacon-enabled mode and a non beacon-enabled mode. In this paper, we evaluate the beacon-enabled based IEEE 802.15.4 MAC due to its simplicity for WSNs applications compared to non beacon-enabled mode [2]. Beacon-enabled mode utilizes the slotted version of CSMA/CA mechanism for contention mechanism and channel access. In CSMA/CA, a long random backoff time causes longer average delay, while a small one gives high collision rate [3–5]. Therefore, this paper examines two main drawbacks of slotted CSMA/CA algorithm; the first problem is that during CSMA/CA mechanism, a node tends to delay for a very limited value of backoff exponent (BE). The probability of collisions when two or more nodes choose the same value of backoff period is high. This insufficient distribution of backoff time causes more collisions among the contending nodes and affects system performance. The second problem is that CSMA/CA updates the contention window length without considering the number of contending nodes in the communication medium. Thus, CSMA/CA is proven to be inefficient in terms of system throughput, reliability and energy efficiency. In this paper, we propose an efficient backoff (EBA-15.4MAC) algorithm based on the IEEE 802.15.4 MAC protocol that resolves the aforementioned problems. To evaluate the performance of EBA-15.4MAC algorithm, the network simulator (NS-2) has been conducted. Simulation results demonstrate that the proposed mechanism significantly improves system throughput and delivery ratio by up to 30 % compared to the existing standard. Furthermore, minimized energy consumption is achieved especially for higher traffic load. The rest of this paper is organized as follows. Section 2 introduces a description of related literature that targeted on modifying the 802.15.4 standard. Section 3 provides a brief overview of the IEEE 802.15.4 MAC protocol. Section 4 analyses and formulates the EBA-15.4MAC algorithm. Sections 5 and 6 gives the performance analysis and the simulation results to validate our proposed model, respectively. Finally, Sect. 7 concludes the paper.

2 Related Work IEEE 802.15.4 is considered a new promising technology for WSNs. Many researchers have been studied the performance of CSMA/CA algorithm and introduced different backoff algorithms. In [6], the authors proposed a memorized backoff scheme (MBS) with the exponential weighted moving average (EWMA) to adjust the contention window size based on the traffic load. In IEEE 802.15.4, collisions occur when two mobile nodes choose the same backoff period due to CWmin especially for high traffic load. Therefore, MBS recorded the CW value for a successful transmission in the previous superframe to the initial value of CW of the current superframe. MSB with EWMA approach choose the accurate initial value of CW. To analyze MBS scheme, the analytical model has been used. To further improve the CSMA/CA, the authors in [7] introduced a new state transition scheme (STS). In STS, they adjusted the backoff exponent (BE) range of IEEE 802.15.4. In this case, the minimum value of BE (macMinBE) will be flexible to change. Hence, they put the value of macMinBE of

123

An Efficient Backoff Algorithm

some nodes to a smaller value and changed it dynamically based on the transmission conditions. In this case, the backoff period of nodes will reduce and the waiting time during the busy detection of CCA will be shortened. This make the node try to sense the channel more frequent and gives a better chance to reduce the packet collision. A delayed backoff (DBA) algorithm [3] resolves the inefficient waiting time of IEEE 802.15.4. In DBA, each contending node has its own backoff time assigned by the coordinator. This is done by applying two different stages; Backoff Period Assignation Stage and Data Transfer Stage. The description of each stage can be found in [3]. Moreover, two mechanisms are proposed in [8]; the enhanced collision resolution (ECR) and the enhanced backoff (EB). The ECR adjusts the backoff exponent depend on the busy results of clear channel assessment (CCA) while the EB is basis on shifting the range of backoff period to reduce redundant backoffs based on the results utilized by the CCA. In this scheme, the expected number of shifting range hasn’t a clearly computation; the average delay may increase due to shifting range of BP. In addition to this mechanism, we can still use the information CCA1 and CCA2 efficiently to detect channel condition. Therefore, the authors in [9] proposed the additional carrier sensing (ACS) algorithm based on the IEEE 802.15.4. ACS detects the channel status whenever the CCA2 detects a busy channel. In this case, ACS can provide accurate information that the busy channel is caused by data or acknowledged packet transmission in the CCA2 detecting. The mathematical model is developed to analyze the ACS algorithm. Furthermore, many new backoff algorithms tries to categorize nodes into groups which are assigned to separate the backoff periods to achieve better performance while maintaining the same level of packet delays, one of them is the Non-Overlapping Binary Exponential Backoff (NO-BEB) algorithm [10]. NO-BEB separates the backoff interval regions into nonoverlapped sub-intervals represented by [2BEi−1 , 2BEi − 1]. It uses the latter half of the interval rather to avoid overlapping with the previous interval. Hence, nodes with unsuccessful channel will be allocated at the non-overlapped areas to choose different random backoff delays to reduce the collision probability among contending nodes. The mathematical model has been developed to analyze the proposed scheme. When focusing on IEEE 802.15.4, we should implement different mechanisms and methods. Recently, many researchers highlighted the study of contention access period (CAP) and the contention free period (CFP) of IEEE 802.15.4. In [11], the authors analyzed the performance of CAP by using the Markov chain models of the node states and the channel states. In addition, an enhanced collisionavoidance MAC protocol has been proposed to improve the system performance of IEEE 802.15.4. In this paper, we mainly focus on the backoff period in this standard. Therefore, we propose an efficient backoff (EBA-15.4MAC) algorithm based on IEEE 802.15.4 beaconenabled mode to enhance the slotted CSMACA with analysis of the MAC layer design to achieve better system performance. We also compare our results against the IEEE 802.15.4 MAC protocol standard using NS-2 simulator.

3 Overview of IEEE 802.15.4 IEEE 802.15.4 defines the characteristics of the PHY and MAC layers for low rate wireless personal area networks (LR-WPANs). Two basic network topologies, star (single-hop) and peer-to-peer (multi-hop), are supported in IEEE 802.15.4 [12]. In the star topology, communications are possible between nodes and PAN coordinator to establish and maintain the transmission. In the peer-to-peer topology, a coordinator is also used and nodes can communicate with any other nodes within its transmission range.

123

Z. Dahham et al. Beacon

Beacon

CAP

CFP

GTS

CSMA/CA

GTS

Inactive

SD= aBaseSuperframeDuration*2SO (Active) BI= aBaseSuperframeDuration*2BO

Fig. 1 IEEE 802.15.4 superframe structure [14]

Moreover, the IEEE 802.15.4 MAC layer devices are classified into full function device (FFDs) and reduced function device (RFDs). An FFD is a complete functional device of IEEE 802.15.4 that supports all MAC layer functions and primitives while the RFD is equipped to support a sub set of them [13]. Also an FFD can be act as a network coordinator or as a network end-device. When FFDs is acting as a coordinator, it can send a beacons; defines as a special synchronization frames which is used then for communication between network services. On the other hand, RFDs is only can be act as end-devices and can interact with a single FFD. 3.1 Superframe Structure The IEEE 802.15.4 network can work in either a beacon-enabled mode or non beaconenabled mode. Superframe structure is imposed in the beacon-enabled mode as shown in Fig. 1. The superframe is bounded by a network beacons; a special synchronization frames send periodically by the coordinator. A superframe begins and ends with the beacon frame. The length of the superframe is called the Beacon Interval (BI) and it is size defined within the Beacon Order (BO) parameter as follows: BI = aBaseSuperframeDuration ∗ 2 B O symbols, 0 ≤ B O ≤ 14 Furthermore, each superframe consists of active and inactive period. During the active period, the sensor nodes communicate with coordinator and enter a low-power state to save energy in the inactive period. The active period is referred to as Superframe Duration (SD), and is defined through the Superframe Order (SO) parameter as follows: S D = aBaseSuperframeDuration ∗ 2 S O symbols, 0 ≤ S O ≤ B O ≤ 14 In our simulation scenario, we used an equal value of SO and BO (i.e., S D = B I ) to keep the superframe active at all times. In addition, the active period (SD) is also divided into three parts; a beacon, a contention access period (CAP), and a contention free period (CFP). Any device wants to access the channel and communicates during the CAP; a slotted CSMA/CA mechanism is used. Meanwhile, the CFP contains a number of Guaranteed Time Slots (GTS), which is locates at the end of the active period. In CFP, communication occurs in a time-division multiple access (TDMA) technique. There is no superframe in the nonbeaconenabled mode, devices are always active and unslotted CSMA/CA algorithm is utilized for channel access.

123


3.2 Slotted CSMA/CA Algorithm The CSMA/CA algorithm is used in both beacon-enabled mode and non beacon-enabled mode. In this paper, we focus on the beacon-enabled mode which is slotted CSMA/CA is used. In the CAP, each node communicates with the coordinator using the slotted CSMA/CA. Transmission can start at the boundaries of units of time called bakoff slots and indicated by aUnitBackoffPeriod = 20 symbols. A node has a packet to transmit, delays for a random value of backoff period (BP) chosen in range [0, 2BE − 1] slots, where BE is the backoff exponent initialized to the value of macMinBE and has the default value of three. The backoff and then the carrier sensing is applied to reduce the probability of collision among the contending nodes. Once the backoff timer expires to reach zero, two CCAs are performed to detect the channel condition and to ensure the channel is clear of activities. If the channel is found idle, the node starts to transmit its data and wait the coordinator to send acknowledgment packet. Conversely, if either CCA1 or CCA2 detects a busy channel, the value of BE and NB will be increased by one. BE and NB have their maximum values which are aMaxBE, and macMaxCSMABackoffs respectively. If BE exceed its maximum value, it will reassign again in range [0, 2BE+1 − 1], while the transmission will fails and the packets will discard if NB reach to macMaxCSMABackoffs. In this case, failure result will declare to the upper layer. The flowchart of the CSMA/CA algorithm is shown in Fig. 2.

4 Description of the EBA-15.4MAC Algorithm In this section, we give a brief introduction of the proposed EBA-15.4MAC algorithm and then we describe in details the principles of the proposed model. 4.1 Introduction of the EBA-15.4MAC Algorithm The motivation behind EBA-15.4MAC is to enhance the performance of IEEE 802.15.4 slotted CSMA/CA. These enhancements should take into consideration the power consumption issues to be as low as possible in order not to affect network lifetime. Therefore, EBA15.4MAC is designed based on two new techniques; firstly, it updates the contention window size based on the probability of collision parameter. As more nodes attempt to access the channel at the same time, the probability of collision increases as well. Therefore, including the probability of collision in the computation of contention window; can adapt the value of CW according to the conditions in the communication medium. The value of CW is update as follows: CWi = maxCW ∗ Pcolli Pcolli = Nc/(Nc + Nr)

(1) (2)

where, CWi is the contention window at time i, maxCW is the maximum contention window allowed by IEEE 802.15.4 (set to 2macMaxBE ), and Pcolli is the probability of collision at time i. At the beginning of EBA-15.4MAC implementation, we assume that the probability of collision is initialized with the value of 0.6 since the node has not yet sent any packet. The probability of collision (Pcolli ) is computed based on the packets that suffer from collisions. After transmitting a number of packets, the node observes the proportion of packets that suffer from collisions and based on that, the node will calculate the Pcolli as shown in (2). Nc and Nr represent the total number of collided packets and the total number of received

123

Z. Dahham et al.

CSMA-CA

Step (1)

NB=0, CW=2

Y

BE = lesser of (2, macMinBE)

Battery life extension? N

BE = macMinBE

Locate backoff period boundary

Step (2)

Step (3)

Delay for random (2 BE -1) UBPs

Perform CCA on backoff period boundary Y

Channel Idle? Step (5) N Step (4)

CW=2, NB=NB+1 BE = min(BE+1, macMaxBE)

CW = CW-1

N N

NB> macMaxCSMABackoffs?

Y

Failure

CW =0?

Y

Success

Fig. 2 Slotted CSMA/CA algorithm

packets observed by the node, respectively. In this case, the contention window size will be updated based on (1) and (2) as we discussed in the next section. Secondly, to increase the efficiency of EBA-15.4MAC, we propose a new scheme to resolve the problem of access collision due to the small value of BE used by CSMA-CA. In this scheme, we allow the nodes to delay for a random number of backoff periods by employing a novel Temporary Backoff (TB) and Next Temporary Backoff (NTB). Hence, the nodes not choose BE randomly as mentioned in the standard but they select TB and NTB within 10–50 % of the actual backoff delay selected randomly by the node. By employing a random selection of TB and NTB values within the whole existing backoff delay, EBA15.4MAC minimizes the chances of access collision; since the probability of two nodes

123


selecting the same random TB value as well as NTB value is very low [15,16]. TB and NTB are calculated as follows. TB = (Backoff ∗ TP)/100 NTB = Backoff − TB

(3) (4)

where TP is the actual percentage of the backoff delay (TP = 10, 20, 30, 40, 50), and Backoff is the value of random delay selected by the node for the backoff. The value of Backoff is Backoff = CWi ∗ UBPs

(5)

By employing these new techniques, EBA-15.4MAC resolves the problem of access collision and significantly improves throughput and delivery ratio, while reducing the power consumption and average packet delay. 4.2 Working Principles of the EBA-15.4MAC Algorithm The EBA-15.4MAC algorithm is used before the transmission of data frames within the contention access period (CAP) of IEEE 802.15.4. The EBA-15.4MAC flowchart is shown in Fig. 3. When the node has a packet to transmit, it first initializes the variables; NB = 0 and maxCW = 2maxBE . In this case, the maximum window size is used to allow the node to delay for an extend period of time. The intention behind that is to reduce the contention among the nodes when the number of nodes increases in the network. Since the node has not yet sent a packet, we assume that the probability of collision (Pcolli ) is initialized with a value of 0.6. This initialization is mandatory so that the node will be able to use Eq. (1) to compute its first contention window. Although 0.6 may seem high, it will not affect the system performance of EBA-15.4MAC since a node will correct and update this value according to the level of collisions observed at the EBA-15.4MAC implementation. In the proposed algorithm’s design, the backoff period has been defined as shown in (5). By including the Pcolli in the computation of contention window, we adapt the backoff period according to the level of collisions detected by the transmitting node. Hence, when the number of collisions increases in the communication medium, the backoff period (Backoff) can be extend for more durations of time. This effectively decreases the contention among nodes and gives better chances for successful transmission. On the other hand, as the level of collision decreases in the network due to lesser number of nodes, nodes tend to delay for shorter backoff duration (Backoff) according to Eq. (5). This reflects in better system performance by reducing packet delay due to short backoff period (Backoff). After setting the appropriate duration of BP (Backoff), the node chooses randomly a unit Temporary Period (TP) which is equal to 10, 20, 30, 40 and 50 in order to delay for a random TB and NTB respectively, where TP represents the actual percentage of the backoff delay. The node then delays for a random value of temporary backoff (TB) within 10–50 % of the backoff delay value selected by node randomly. In this case, we reduce the probability of choosing the same range of BE due to inclusion of TB which results in minimize the number of collisions. TB is computed as shown in (3). We notice that the whole operation of defining the backoff period and then the delay for random number of BP differs from that of the existing CSMA/CA mechanism. The idea here is that nodes have a lower probability of choosing, randomly, the same value of TB and later, NTB. Hence, the collision rate decreased over the network.

123

Z. Dahham et al. EBA-15.4MAC

NB=0 maxCW = 2 macMaxBE

Calculate Pcolli

Backoff = CWi * UBPs

Random TP

Delay for a random TB

Perform CCA on backoff period boundary Y Channel idle? N Delay for a random NTB Success Perform CCA on backoff period boundary

Channel idle? N

Y

NB=NB+1

N

NB> macMaxCSMABackoffs? Y Failure

Fig. 3 The flowchart of the EBA-15.4MAC algorithm

As the node completes the TB delay, it performs the first Clear Channel Assessment denoted as CCA1 to detect the channel’s condition. If the channel is idle of the whole CCA1 period, the node sends its data directly to the coordinator otherwise the node delays for a

123


random number of the next temporary backoff (NTB) according to (4). Upon completing the random value of the NTB delay, the node once again performs CCA2 for data transmission. If the channel is assessed to be free, transmission to the destination resumes otherwise, the node performs the backoff process by increasing the value of NB by one [14]. If NB reaches to its maximum value, that is greater than macMaxCSMABackoffs, the transmission terminates with channel access failure status, otherwise the contention window will be updated according to the level of collision detected by node (1). Finally, the method of updating the contention window is explained as follows. When the level of collision increases in the network, the probability of collision (Pcolli ) will increase as well; causing an increase in the contention window size according to Eq. (1). In this case, nodes tend to backoff for extended time as demonstrated in (5). This will decrease the contention among nodes and give better system performance of successful packet transmissions since the probability of two nodes select the random TB and NTB value is very low. On the other hand, as the level of collision decreases in wireless medium, the backoff duration (Backoff) will be shorter (5). This reflects in better power efficiency and improves system throughput. 5 Performance Analysis In this section, we analyze the performance of the proposed EBA-15.4MAC algorithm based on beacon-enabled IEEE 802.15.4 slotted CSMA/CA. In our simulation, we consider a single hop star network topology consisting of a coordinator and N sensor nodes. Each node is placed at a distance of 10 m far from the coordinator. The traffic is periodically generated in the sensor nodes according to Constant Bit Rate (CBR) traffic flows with one way communication from the node to the coordinator. We assume that each data packet has a fixed length of 70 bytes. The duration of a time slot is equal to a UBP. We also assume that the system consist of active period without CFP or inactive period. By using only the active period, the nodes can follow the slotted CSMA-CA mechanism for channel access. We also considered that all nodes work within the carrier sensing of each other in order to avoid any interruption in a current transmission by other nodes. Table 1 summarizes the parameters of the simulation model.

Table 1 The parameters of the simulation model

Parameter

Value

Topology

Star

Sensor nodes (N )

15

Packet length

70 bytes

Routing type

AODV

Traffic load

0–3

Data rate

20 kbps

Simulation time

1000 s

BO = SO

3

aBaseSlotDuration

60 symbols

macMinBE

3

macMaxBE

8

aUnitBackoffPeriod (UBP)

20 symbols

maxCW

2maxBE

123

Z. Dahham et al.

6 Simulation Results In this section, we compare the results of EBA-15.4MAC against that of IEEE 802.15. Two different simulation scenarios are conducted using NS-2 simulation. In the first scenario, we simulate the proposed algorithm where 15 sensor nodes with one PAN coordinator are deployed and 8 CBR traffic flows. The range of traffic load has been chosen from 0 to 3 packet/s. This provides all three necessary areas of packets servicing, congestion area and the steady state part. In addition, the second scenario considers a network with 35 active nodes includes one PAN coordinator which is located at the center of the network and all end devices located around it. We further compare the results of the proposed algorithm with IEEE 802.15.4 according to the network throughput, power consumption, average packet delay, delivery ratio and power efficiency. The results show the performance of each metric of a device under a saturation condition, i.e., every device always has a packet to transmit. The results are presented with a 95 % confident level. Figure 4 shows the results of the performance measurement versus traffic load. Simulation results of the proposed EBA-15.4MAC algorithm are compared with the IEEE 802.15.4. In Fig. 4, we see the results obtained by the EBA-15.4MAC algorithm for throughput and

(a)

8000

Throughput (bps)

7000 6000 5000 4000 3000 2000 1000

IEEE 802.15.4 EBA-15.4MAC

0

(b)

100

Delivery ratio (Percent)

Traffic load (pkts/sec)

80 60 40 20

IEEE 802.15.4 EBA-15.4MAC

0

Traffic load (pkts/sec) Fig. 4 Throughput and delivery ratio versus traffic load with number of nodes = 15. a Throughput versus traffic load; b delivery ratio versus traffic load

123


delivery ratio are better than the IEEE 802.15.4 especially at higher traffic load. This means that the probability of two nodes selecting the same backff period is low due to inclusion the Temporary backoff (TB) and Next Temporary Backoff (NTB). Thus, our proposed algorithm resolves the backoff problem and minimizes the level of collision. Figure 5 shows the performance of IEEE 802.15.4 and EBA-15.4MAC in terms of power consumption, average delay and power efficiency, respectively. In Fig. 5a, both 802.15.4 and EBA-15.4MAC show similar performance and consumed the same level of power for low traffic loads as they work within the congested area and the nodes sent data at much acceptable rates. However, at a higher traffic load, the IEEE 802.15.4 illustrates a jump in the amount of energy dissipation. This arises as a result of network congestion that can be marked by a higher number of packet drops. It is a common problem of all CSMA-CA based MAC protocols when the traffic in a given network increases; the probability of two nodes choosing the same backoff periods is high. Hence, this leads to a packet collision that causes a higher increase in energy consumption. However, by applying our proposed algorithm, the power consumption of sensor nodes will not increase rather; it becomes stable and less than that of using the traditional CSMA/CA. This improvement in power performance is attributed to the reduction in the probability of choosing the same number of backoff slots. Figure 5b plots the average delay for IEEE 802.15.4 and the EBA-15.4MAC under different traffic loads. As seen earlier, all the previous graphs show similar performances at low traffic rates since the network can service more packets and the transmission to the destination occurs successfully. However, at higher traffic loads, packets have to wait longer in order to be serviced by the network. This will bring more collisions with a lot of packet drops. In IEEE 802.15.4, a sudden increase in the amount of packet delay at 1.2pps traffic load will occur while in EBA-15.4MAC, this increase is postponed until 1.5pps traffic load. This is because a node in 802.15.4 allocates small values of backoff exponents and delays for this limited random number indicated in the range of backoff periods. This method increases the number of packet drops and causes more collisions. Our proposed algorithm decreases the number of packet drops by pushing the congested area further till 1.5pps traffic rate due to the selection of higher values for backoff exponents. Figure 5c shows the comparison between IEEE 802.15.4 and EBA-15.4MAC in terms of their power efficiency. It is defined as the ratio of throughput and energy consumption [8]. We can observe from the Fig. 5c that the EBA-15.4MAC achieves higher power efficiency since the amount of energy used by the EBA-15.4MAC algorithm is lesser than that of the IEEE 802.15.4 standard. In addition, the EBA-15.4MAC increases the amount of system throughput rapidly compared to the original standard. At a higher traffic load and after network congestion, the amount of throughput and energy consumption remains constant. Therefore, we used the same pattern of these two parameters to evaluate the performance of power efficiency. Figure 6 shows the results of the IEEE 802.15.4 and EBA-15.4MAC versus the number of active node. In Fig. 6a, we see the effectiveness of our proposed algorithm in terms of throughput compared with that of IEEE 802.15.4. At low number of nodes—less than 10 nodes—the IEEE 802.15.4 and EBA-15.4MAC algorithm have the same amounts of throughput. However, when the number of nodes increase in the network, the throughput drops drastically due to copying the BE’s value of nodes that sends their data successfully. The rest of the nodes will try to access the channel using the same value of BE. Hence, the level of collisions will increase in the network causing degradation in the throughput of 802.15.4 for a higher number of active nodes. In the case of the proposed algorithm; the throughput of EBA-15.4MAC is better than that of the 802.15.4. The main reason is that, EBA-15.4MAC updates the backoff duration according to the level of collision detected. Besides, the nodes

123

Z. Dahham et al.

(a) 0.45 Energy consumed [%]

0.4 0.35 0.3 0.25 0.2 0.15 0.1

IEEE 802.15.4

0.05

EBA-15.4MAC

0


(b) 120

Delay (sec)

100 80 60 40 IEEE 802.15.4

20

EBA-15.4MAC

0


(c)

180

Power Efficiency(bps/J)

160 140 120 100 80 60 40

IEEE 802.15.4

20

EBA-15.4MAC

0

Traffic load (pkts/sec) Fig. 5 Energy consumption, average packet delay and power efficiency versus traffic load with number of nodes = 15. a Energy consumption versus traffic load; b average packet delay versus traffic load; c power efficiency versus traffic load

from the inclusion of both TB and NTB perform CCA at the temporary backoff. Therefore, the nodes most of the time avoid collision and strive to improve the network throughput. In Fig. 6b, we also see that the average delay of the EBA-15.4MAC is smaller than the IEEE 802.15.4. The results indicate that both curves perform the same when the number of

123


(a) 700 IEEE 802.15.4

Throughput (bps)

600

EBA-15.4MAC

500 400 300 200 100 0

5

10

15

20

25

30

35

30

35

Number of active nodes

Delay (sec)

(b) 450 400

IEEE 802.15.4

350

EBA-15.4MAC

300 250 200 150 100 50 0

5

10

15

20

25

Number of active nodes Fig. 6 Throughput and average delay versus number of nodes with traffic load = 1.5. a Throughput versus number of active nodes; b average delay versus number of active nodes

active nodes not exceeds 10 nodes but when the number of nodes increases in the network, we can see the difference between the two methods. This reason behind is that when the number of contending nodes is low, the probability of choosing the identical value of backoff exponent is low. Hence, the packets consume less time to reach the destination. On the other hand, as the number of nodes increase, the blind backoff process, which is used by 802.15.4, causes a longer average delay. When a node wants to transmit the data, it cannot perform a CCA until the backoff process complete. Hence, it will cause longer delay. The graph clearly shows that, in comparison to IEEE 802.15.4, the proposed algorithm is capable of achieving the smallest values of average delay. This is due to the fact that during the backoff procedure, the nodes perform CCA at the temporary backoff level because of the inclusion of TB and NTB. In this case, the node does not have to wait until the completion of the whole random backoff delay but when the TB time expires, the node can perform CCA. Due to this reason, the loss of packet time is lessened. Figure 7 shows the performance of IEEE 802.15.4 and EBA-15.4MAC in terms of delivery ratio and power efficiency under a different number of active devices. Apparently, at a low number of active nodes, the two schemes are able to achieve higher delivery ratio and power efficiency compared with 802.15.4. This means that most of the transmitted packets are capable of reaching their destination successfully due to the fact that the probability of two

123

Z. Dahham et al.

(a) 100 IEEE 802.15.4 EBA-15.4MAC

Delivery Ratio [%]

80

60

40

20

0

5

10

15

20

25

30

35

Number of active nodes

Power Efficiency (bps/j)

(b) 180 160

IEEE 802.15.4

140

EBA-15.4MAC

120 100 80 60 40 20 0

5

10

15

20

25

30

35

Number of active nodes Fig. 7 Delivery ratio and power efficiency versus number of nodes with traffic load = 1.5. a Delivery ratio versus number of active nodes; b power efficiency versus number of active nodes

nodes selecting the same backoff interval is low. As a consequence, the level of collisions would be low. On the other hand, as the number of nodes gets bigger, the delivery ratio starts to decrease. This is because when the number of nodes increases in the network, packet drops start to happen and most of the packets will lost. In addition, packet collisions increase because of the lower value of BE and therefore, we will need an efficient distribution of backoff time among the nodes. With that said; EBA-15.4MAC can achieve promising results and improve delivery ratio and power efficiency especially at 10 nodes and beyond compared to the standard because of the inclusion of the temporary backoff and the next temporary backoff as we have mentioned beforehand. Again, the EBA-15.4MAC achieves higher and more stable delivery ratio not only at a higher number of active nodes but also at times when the network may experience high traffic loads. With that, the probability of a successful packet transmission is significantly improved.

7 Conclusions In this paper, the EBA-15.4MAC based on the IEEE 802.15.4 beacon-enabled mode is proposed. It uses two important schemes. Firstly, it is the idea of controlling the contention

123


window size and the backoff duration according to the level of collision detected over the communication medium. In this case, when the shared medium expects a high level of collision, nodes tend to backoff for an extended duration of time to decrease the number of packet sending. On the other hand, as the level of collision starts to decrease, an idle medium is avoided by decreasing the backoff time. Secondly, EBA-15.4 MAC allow the nodes to delay for random backoff periods using Temporary Backoff (TB) and Next Temporary Backoff (NTB) that are chosen randomly from (10 to 50)% of the total backoff delay value. With this, the EBA-15.4MAC eliminates the inefficient range of backoff exponents then reduced collisions among the contending nodes. By applying the proposed techniques, possible substantial improvements are shown in the overall system performance. Finally, simulation results confirmed our expectations and demonstrated that the proposed EBA-15.4MAC algorithm significantly outperforms the original IEEE 802.15.4 MAC protocol.

References 1. Baronti, P., Pillai, P., Chook, V. W. C., Chessa, S., Gotta, A., & Hu, Y. F. (2007). Wireless sensor networks: A survey on the state of the art and the 802.15.4 and ZigBee standards. Computer CommunicationsElsevier, 30, 1655–1695. 2. Koubaa, A., Alves, M., & Tovar, E. (2006). A comprehensive simulation study of slotted CSMA/CA for IEEE 802.15.4 wireless sensor networks. In IEEE international workshop on factory communication systems (pp. 183–192). 3. Bih-Hwang, L., & Huai-Kuei, W. (2007). A delayed backoff algorithm for IEEE 802.15.4 beaconenabled LR-WPAN. In 6th International conference on information, communications & signal processing (pp. 1–4). 4. Chi-Ming, W., Ruei-Lung, L., & Lai, I. T. (2010). An enhanced carrier sensing algorithm for IEEE 802.15.4 low-rate wireless sensor networks. In IEEE symposium on industrial electronics & applications (ISIEA) (pp. 10–15). 5. Wong, C.-M., & Lee, B.-H. (2012). An Improvement of Slotted CSMA/CA Algorithm in IEEE 802.15.4 Medium Access Layer. Wireless Personal Communications, 63, 807–822. 6. Ai-Chun, P., & Hsueh-Wen, T. (2004). Dynamic backoff for wireless personal networks. IEEE Global Telecommunications Conference (GLOBECOM), 3, 1580–1584. 7. Jeong-Gil, K., Yong-Hyun, C., & Hyogon, K. (2006). Performance evaluation of IEEE 802.15.4 MAC with different backoff ranges in wireless sensor networks. In 10th IEEE Singapore international conference on communication systems (ICCS) (pp. 1–5). 8. Jae Yeol, H., Kim, T. H., Hong Seong, P., Sunghyun, C., & Wook Hyun, K. (2007). An enhanced CSMACA algorithm for IEEE 802.15.4 LR-WPANs. IEEE Communications Letters, 11, 461–463. 9. Wong, C.-M., & Lee, B.-H. (2012). An improvement of slotted CSMA/CA algorithm in IEEE 802.15.4 medium access layer. Wireless Personal Communications, 63, 807–822. 10. Seung-Youn, L., Youn-Soon, S., Jong-Suk, A., & Kang-Woo, L. (2009). Performance analysis of a nonoverlapping binary exponential backoff algorithm over IEEE 802.15.4. In Proceedings of the 4th international conference on ubiquitous information technologies & applications (ICUT) (pp. 1–5). 11. Wang, F., Li, D., & Zhao, Y. (2011). On analysis of the contention access period of IEEE 802.15.4 MAC and its improvement. Wireless Personal Communications, 65, 955–975. 12. Anastasi, G., Conti, M., & Di Francesco, M. (2011). A comprehensive analysis of the MAC unreliability problem in IEEE 802.15.4 wireless sensor networks. IEEE Transactions on Industrial Informatics, 7, 52–65. 13. Lee, B. H., & Wu, H. K. (2008). Study on backoff algorithm for IEEE 802.15. 4 LR-WPAN. In 22nd International conference on advanced information networking and applications (AINA) (pp. 403–409). 14. IEEE Standard for Information Technology, Part 15.4; Wireless Medium Access Control (MAC) and Physical Layer (PHY) Specifications for Low-Rate Wireless Personal Area Networks (LR-WPANs), IEEE Computer Society, 2006. 15. Woo, S., Park, W., Ahn, S., An, S., & Kim, D. (2008). Knowledge-based exponential backoff scheme in IEEE 802.15.4 MAC. Information networking. Towards ubiquitous networking and services (pp. 435– 444). 16. Ye, W., Silva, F., & Heidemann, J. (2006). Ultra-low duty cycle MAC with scheduled channel polling. In 4th International conference on embedded networked sensor systems (pp. 321–334).

123

Z. Dahham et al.

Author Biographies Zahraa Dahham received the B.Sc. degree in electronics engineering from the Mosul University, Mosul, Iraq, in 2008. She is currently a master candidate in the Wireless Laboratory of Computer and Communication System Engineering, Faculty of Engineering, University Putra Malaysia, Malaysia. Her current research interests include the area of wireless communications and WSNs.

Aduwati Sali received the B.Eng. degree (Hons) in electrical and electronics engineering from the University of Edinburgh, UK, in 1999. She received her M.Sc. in communications and network engineering from the University Putra Malaysia, Malaysia, in 2002 and her Ph.D. in mobile satellite communications from the University of Surrey, UK, in 2009. Currently, she is a Senior Lecturer in the Dept of Computer and Communication Systems Engineering, Faculty of Engineering, University Putra Malaysia, Malaysia. Her research interests include mobile and satellite communications, WSNs, heterogeneous networks, green radio resource management, cognitive radio, adaptive techniques, cross-layer approach, reliable multicast, disaster management and preparedness.

Borhanuddin M. Ali received his B.Sc. degree (Hons) in electrical and electronics engineering from the Loughborough University of Technology, UK, in 1979, and the M.Sc. and Ph.D. degrees from the University of Wales, Cardiff, UK, in 1981 and 1985, respectively. He was a director of Institute of Multimedia and Software in Malaysia. He founded the national networking testbed project code named Teman, and became Chairman of the MYREN Research Community in 2002. He is a Senior Member of IEEE and a member of IET and a Chartered Engineer. Currently, he is a professor in the computer and communication system engineering, University Putra Malaysia, Malaysia. His research interests include broadband and high speed networks, broadband wireless access and optical networks.

123