Power-Optimization Method for the Implementation

GESTS Int’l Trans. Computer Science and Engr., Vol.60, No.1

1

Power-Optimization Method for the Implementation of IEEE 802.11 Multi-Hop Networks L. N. Gumbi1 , MO. Odhiambo2 ,and D. Chatelain1 1 French South African Technical Institute in Electronics Tshwane University of Technology, Private Bag X680 Pre ria 0001, South Africa {lucas.gumbi, Damien.chatelain}@fsatie.ac.za 2 Department of Electrical & Mining Engineering, university of South Africa, Private Bag X6 Florida 1710, South Africa [email protected]

Abstract: In recent years the telecommunications society has witnessed an upsurge of interest in infrastructure-based multi-hop communication. The fundamental reason being that multi-hop communication provides benefits such as coverage e tensions, combating of shadowing at high radio frequencies and reduced infrastructure deployment costs. It is envisioned that the WLAN technology will likely be merged with future wireless broadband technologies such as 3G, for the provisioning of pervasive and user seamless communication, however, WLANs are limited in range. One standard approach to extending the range of WLANs is to use multi-hop communication systems, but, in IEEE 802.11, multi-hop communication systems are not optimized in terms o ower consumption. This paper presents a power-efficient multi-hop communication algorithm (PMCA) for the implementation of IEEE 802.11 multi-hop networks. The proposed scheme estimates the number of nodes, transmit power level and the transmitter-receiver separation distance based on transmission distance, receiver sensitivity, energy consumption and infrastru ture costs. Simulation results show that our algorithm significantly optimizes power consumption as compared to the conventional multi-hop communication.

Key words: Relays, Power Consumption, IEEE 802.11b Networks, Multi -hop Communication.

I INTRODUCTION

T

here has been a phenomenal spate of interest in multi -hop communication in recent years. This is indicated by the seed concept in 3GPP, mesh networks in IEEE 802.16, and cov age extensions in HiperLAN/2 through relays [3]. A multi hop wireless network is one in which a packet may have to traverse multiple consec tive wireless links in order to reach its destination [4]. Over the years multi -hop communication has manifested itself under numerous forms, such as Packet Radio Networks developed several decades ago for military applications, and more recently ad hoc networks which refer to a collection of hosts communicating over a wireless chan . Multi-hop communication provides benefits such as range extension, combating of shadowing at high frequencies, network partitioning avoidance and reduced infrastructure de oyment costs. The IEEE 802.11 WLANs are becoming increasingly more p pular, as they are envisioned as the cornerstone of future wireless broadband communications, which are character by very high data rates. The merging of such heterogeneous technologies may form a network with an increased syst m capacity and hence improved system performance. The resulting network may use WLAN techniques for peer -to-peer communication or serve as relay networks to extend the range and increase the capacity of future wireless broadband net orks, which provides links to the wired Internet [1]. WLANs were fundamentally designed to provide broadband wireless a ess to home and office users, however that notion has changed now, since it is clear that they will play an integral part in future wireless broadband networks. WLANs use traditional LAN technology with a wireless interface, and provide rela ively high-speed data communication (up-to 11Mbps for IEEE 802.11b) in small areas such as a building or an office [2]. The w eless connection allows users to roam in a bounded area while still connected to the networks. Even though IEEE 802.11 families of WLAN provide a low cost broadband wireless access, they have a limited transmission range. One way to extend the range is to use multi-hop communication. The typical way of implementing multi -hop communication in static networks, is to randomly select node transmit power levels (normally maximum transmit power) that can link the nodes together to cover a specific range (conventional multi -hop system). This mechanism result in higher consumption of transmit pow and energy, and a higher probability of interference. Power is a scarce resource in wireless networks, consequently we cannot afford its wastage. In IEEE 802.11, the main standard approach used for tr it power optimization is varying the transmit power level to the minimum level that still achieves correct reception of packets despite the intervening interference [13]. This approach is realized through TPC (transmit power control) schemes veloped within the IEEE 802.11 framework [4][14][15]. However, the work done by [14][15][16] indicates that the TPC s mes have the following disadvantages: 1. 2. 3.

Most TPC schemes use different transmit powers on different nodes, this creates asymmetric links that result in more collisions and consequently network throughput degradation. Applying TPC schemes on real multi -hop networks raises unexpected issues not revealed during simulations.

Most TPC schemes cannot be implemented efficiently on urrent wireless IEEE 802.11 devices because of hardware, driver and operating systems issues. In essence, the existing IEEE 802.11 devices do not fulfil the requirements of transmit power control algorithms.

ⓒGESTS-Jun.2010

2

Power-Optimization Method for the Implementation of IEEE …

The latter TPC issues clearly indicate that there is a need for a practically viable, power efficient and cost effective solution to power optimisation problems of IEEE 802.11 multi-hop networks. This paper presents an efficient power consumption multihop communication algorithm for IEEE 802.11b. Unlike t ventional multi -hop system and the latter TPC schemes, our proposed scheme estimate the number of nodes, transmit power level and the T-R separation distance based on transmission range, receiver sensitivity, energy consumption and inf astructure costs to optimize transmit power and subsequently energy consumption. In this paper we consider the IEEE 802.11b standard, w h supports 1Mbps, 2Mbps, 5.5Mbps and 11Mbps data rates. The standard is chosen for its popularity and widespread use in the WLAN infrastructure. The rest of the paper is organized as follows. In Sect on II, we analyze multi-hop communication with regard to transmit power. Section III presents the modeling of the IEEE 802.11b performance metrics, i.e. received signal strength (RSS), Signal to-noise ratio (SNR), Bit Error Rate (BER), throughput an the wireless Channel. The proposed algorithm is presented in Section IV. In Section V we evaluate the performance of the algorithm. Finally, Section VI concludes the paper.

II RELAY MULTI-HOP COMMUNICATION In this section we analyze the relationship between mu ti -hop communication and transmit power. Our objective is to show that transmit power is significantly reduced when increasing the number of hops from source to destination. Multi -hop communication is illustrated in Figure 1 using two hop . (a)

A

(b)

A

D

da

R

B

db

B

Figure 1: Two hop Multi-hop communication From Figure 1, Node A and B represent the transmitter Tx) and receiver (Rx) situated a distance D apart, and R represent a relay node. The link between Tx and Rx in terms of transmit and received power is mathematically expressed as: 2

P ⎛ l ⎞ P Rx = ⎜ ⎟ PTx = Tx LP ⎝ 4p D ⎠

(1)

2

where L = ⎛⎜ 4 p D ⎞ is the free space path loss between two isotropic antennas. P ⎝ l ⎠ Rearranging equation (1) resolves into 2

⎛ 4p ⎞ PTx = PRx ⎜ ⎟ D 2 ⎝ l ⎠

(2)

2

Let ⎛⎜ 4p ⎞⎟ = K

⎝ l ⎠

where K is a cons tant, ? = c / f, c is speed of light and

f transmission frequency.

Equation (2) illustrates that transmit power in a single hop indicated by Figure 1(a) is expressed as: (3)

PTx = PRx KD 2 Two hop communication indicated by Figure 1(b) is math matically expressed as:

PTxA + PTxR = PRx Kd a + PRx Kd b D where d a = d b = , n being the No of hops n 2

2

(4)

From equation (4), we can now generally define multi -hop communication as: 2 2 ⎧⎪ ⎫⎪ ⎛ D⎞ ⎛ D⎞ PTx = ⎨n ⋅ PRx K⎜ ⎟ ; ∀1 < n < ∞ | 0 ≤ PRx K⎜ ⎟ < 20dBm⎬ ⎪⎩ ⎝n⎠ ⎝ n⎠ ⎪⎭ n

ⓒGESTS-Jun.2010

(5)

GESTS Int’l Trans. Computer Science and Engr., Vol.60, No.1

3

Figure 2: Exponential decay of Tx power with increasing nodes. Equation (5) indicates that, by increasing the number of nodes within a source-to -destination transmission, transmit power is significantly reduced and hence power consumption is s ificantly minimized within the communication system. A plot shown in Figure 2 illustrates this observation for a Wi -Fi communication system. Figure 2 is plotted using equ tion (5) with n = 20 and D = 1100m. It is shown in Figure 2 that transmit power decays exponentially with the increasing number of hops. In general the decay is defined as: -

a = e− b

(6)

Where a is the transmit power, and b is the number of hops. It is also shown from Figure 2 that the minimum number of hops required to establish a communication link is 5 s each node can only use a maximum power of 100mW as indicated by equation (5), however if more than 5 hops are utili ed, transmit power is significantly reduced and power consumption is optimized. These observations indicate that in multi -hop communication, there exists a threshold point that enables the implementation of power -efficient multi -hop communication systems with optimized system costs. The main objective of our proposed algorithm is to find this threshold point. The proposed power -efficient multi -hop algorithm is discussed in Section IV.

III NETWORK P ERFORMANCE METRICS Throughput, RSS, SNR and BER are used in IEEE 802.11 b ed wireless communication as the basic metrics that determines the performance of the system. In this paper we will u these performance metrics to validate the results achieved by our algorithm. A. RSS and SNR SNR is defined as the ratio of Energy per bit E b to noise power spectral density N o . In IEEE 802.11b Eb / N o is modified as shown in (12), and hence the SNR is expressed as in [5]: ⎛E SNR = 10 log ⎜⎜ c ⎝ No

where

Ec

⎞ ⎟⎟ ⎠

(7)

is the energy per chip

The SNR is defined as in (7) due to a constant chip rate of 11Mchips/s used in IEEE 802.11b. The Direct Sequence Spread Spectrum (DSSS) PHY Layer modulates each transmitted bit with an 11-bit Barker code compelling the chip rate to be fixed at 11Mchips/s. This technique decreases noise energy i side the receiver bandwidth, and hence probability of error is also reduced. The Additive White Gaussian Noise (AWGN) at the receiv (802.11 devices) without the Noise Factor is expressed as:

ⓒGESTS-Jun.2010

4

Power-Optimization Method for the Implementation of IEEE …

N o = kTB where

(8) - 23

k = Boltzmann constant = 1.38 * 10 J/°K/Hz T = ambient temperature = 20 °C = 293°K B = bandwidth of receiver = 18 MHz

Therefore SNR =

RSS

(9)

No

where RSS is the received signal strength expresse d as in equation (1) B. Bit Error Probability The BER (Bit Error Probability) is used as the standard measure of the quality of digital signals in wireless communication systems. It indicates the ratio of the number of bits received in error vs the total number of bits sent. The higher the bit error probability, the poor the reliability of a communication system. For a given SNR, the bit error probability can be easily attained for the IEEE 802.11b modulation schemes (DBPS K, and CCK).

Figure 3: PLCP Protocol Data Unit (PPDU) Format [7] The IEEE 802.11 frame format is shown in Figure 3. The Physical Layer Convergence Procedure (PLCP) Preamble and the PLCP Header are transmitted using DBPSK modulation at a transmission rate of 1Mbps and the MPDU (MAC Protocol Data Unit) is transmitted on the chosen rate among 1, 2, 5.5 and 11Mbps [8]. The frame error probability is mathematically expressed as:

(

)(

Pf rame_ error =1 − 1− Pphy_ error ⋅ 1 − PMAC_ error where

)

Pphy _ error is the physical layer overhead error probability and PMAC _ error

(9) the MPDU error probability.

Pphy _ error = 1 − (1 − Pbit _ error1 )

24⋅8

where

(10)

Pbit _ error1 is the bit error probability when transmitting physical layer overhead.

PMAC _ error = 1 − (1 − Pbit _ error 2 )(28+ MSDU)⋅8 where

Pbit _ error 2 is the bit error probability when transmitting the MPDU.

To express Pb it _ e rro r 1 and

Pbit _ error 2 , E c / N o is derived from E b / N o using the 11 -bit Barker spreading sequence. The

Barker sequence modifies E b

⎛ Eb ⎜ ⎜N ⎝ o The

(11)

/ N o to the following equation.

⎞ E ⎟ = 11 ⋅ c = 11 ⋅ SNR ⎟ No ⎠ DBPSK

(12)

Pbit _ error1 and Pbit _ error 2 is now derived from (12) for IEEE 802.11b modulation s

For 1 Mbps DBPSK,

es.

Pb it _ e rro r 1 is expressed as:

(

Pbit _ error1 = Q 11⋅ SNR

)

This equation applies to every operational mode since

ⓒGESTS-Jun.2010

(13) is also used for physical layer overhead information transmission.

GESTS Int’l Trans. Computer Science and Engr., Vol.60, No.1

For 2 Mbps DQPSK,

Pbit _ error 2 is expressed as:

(

Pb it _ e rro r 2 = Q 5.5 ⋅ SNR For 5.5 Mbps CCK5. 5 ,

)

(14)

Pbit _ error 2 is expressed as:

(

Pbit _ error 2 = min 1 − (1 − SER) , 0.5 where

(

0.25

) (

SER = 14 ⋅ Q 8 ⋅ SNR + Q 16 ⋅ SNR

For 11 Mbps CCK1 1,

)

Pbit _ error 2 is expressed as:

(

)

(15)

)

Pb it _ e rro r 2 = min 1 − (1 − SER ) ,0.5 where

5

1/ 8

(16)

(

) ( ) + 174 ⋅ Q ( 8 ⋅ SNR )+ 16 ⋅ Q ( 10 ⋅ SNR )+ 24 ⋅ Q ( 12 ⋅ SNR )+ Q ( 16 ⋅ SNR ) SER = 14 ⋅ Q 8 ⋅ SNR + Q 16 ⋅ SNR

The Q function is defined as the area under the tail o the Gaussian probability density function with zero mean and unit variance. It is expressed as:

Q( x ) =

1 2p

∞

∫e

2

t 2

dt

(17)

x

C. IEEE 802.11b Throughput Analysis

In this subsection we present an analysis of the IEEE 802.11b throughp t using similar approach as in [7] and [8]. Throughput is analyzed under Distributed Coordination Function (DCF) MAC layer scheme illustrated in Figure 4. In this paper we consider the same parameters, assumptions and conditions considered in [7] and [8].

Figure 4: IEEE 802.11 MAC Layer DCF Operation To model the throughput we need to calculate the average time to transmit a single frame. The average transmission time for a single frame is dependent on the following The time taken to successfully transmit a single frame 1. The time taken between two consecutive frame transmissions if there exists a failed frame transmission 2. As shown in Figure 4, the time taken to successfully t ansmit a frame is defined as:

TS (i) = DIFS+ Tb a c k o ff(i ) + Td a ta + SIFS+ Ta c k [ms]

(18)

where i = the number of consecutive unsuccessful transmissions. DIFS = Distributed Inter Frame Space period SIFS = Short Inter Frame Space period Tbackof f (i) = Backoff Time Interval

ⓒGESTS-Jun.2010

6

Power-Optimization Method for the Implementation of IEEE …

Tdata = Data frame transmission duration Tack = Acknowledgement frame transmission duration. Tdata and Tack are defined as: ⎛ 14 ⋅ 8 ⎞ ⎛ (28 + MSDU)*8 ⎞ (19) Tack = tPhy PLCPpreamb le + tPhy PLCPheader + ⎜⎜ ms ms [ ] txr Mbps ⎝ ⎠ ⎝ txr [Mbps ]⎠

Tdata = tPhyPLCPpreamble + tPhyPLCPheader + ⎜⎜

where txr = Transmission rate (1,2,5.5 or 11Mbps) tPhy PLCPpreamb le and tPhyPLCPheader = Physical layer MAC Service Data Unit (MSDU) = Information

(19)

overhead information. passed from IP layer to MAC layer.

Tback off (i ) is expressed as: ⎛ 2 i ( CW min + 1) − 1 ⎞ ⎟⎟ * S LOT [ms ]0 ≤ i < 6 2 ⎝ ⎠

(20)

⎛ CW max ⎞ * SLOT [ms ]i ≥ 6 2 ⎝ ⎠

(21)

Tbackoff ( i ) = ⎜⎜

Or T b a cko ff ( i ) = ⎜

Secondly, if the frame transmission fails then the time between two consecutive frame transmissions is defined as: Tf (i) = DIFS + Tback off(i) + Tdata + (SIFS + Tack + SLOT)ms By using (18) and (22) we derive the average transmission time for a single me as: +∞ ⎡ i −1 ⎤ T fra m e = T S ( 0) + (1 − p )p i ⎢ T f ( j ) + T S (i) ⎥ [ms ] i =1 ⎣ j=0 ⎦

∑

∑

(22)

(23)

Throughput is then finally modeled as: Throughput

⎛ MSDU * 8 ⎞ = ⎜ Mbps ⎜ T fra m e ⎝ ⎠

(24)

D. Channel Modeling In a wireless environment, the RSS is largely influenc d by transmit power, large-scale path loss and sma ll-scale multi-path fading. Path loss determines the mean RSS as a functio of distance. Multi-path fading is caused by the superposition of multiple in-phase and out -of -phase copies of the original transmitted signal. It causes rapid fluctuations in the RSS over very short time scales [9]. Since we are dealing with fixed nodes we do not consider multi-path fading in this paper, because there is relatively no movement between the sender and the r eiver. Path loss is a fundamental and very important factor in the design or simulations of wireless systems. It i e link between the transmitter and the receiver. There are several well -established mathematical models for path loss. In this paper we consider the log-distance path loss model described in [10].

where

d0

⎛ d ⎞ PL( dB) = PL( d 0 ) + 10 ⋅ n ⋅ log ⎜⎜ ⎝ d0 ⎠ is the close-in reference distance, d the distance between the transmitter and the receiver, and

(25)

n the path loss

exponent defined in [10].

IV P OWER EFFICIENT MULTI-HOP ALGORITHM In this section, we present the implementation of our proposed algorithm. The analysis presented in section II, has shown that increasing the number of hops within a source to destination transmission reduces power consum . However, it was also indicated that increasing the number of hops result in increased infrastructure costs, which in turn increases the overall system costs. Thus, there exists a point that optimizes both power consumption and system costs. Our pro osed algorithm is designed to establish this threshold point. Its principle is to estimate the number of nodes, transmit power level and the T-R separation distance of nodes based on transmission distance, receiver sensitivity, energy consumption and infrastructure costs.

ⓒ GESTS-Jun.2010

GESTS Int’l Trans. Computer Science and Engr., Vol.60, No.1

7

A. Proposed Algorithm The basic idea of our power efficient multi -hop algorithm is to establish a threshold point that d mines if a multi -hop communication is optimized in terms of power consumption and infrastructure costs. The threshold point is expressed as:

PCij = { (Pi − Pj )⋅ Dij (t, p) ⋅ EC | Pi ≠ Pj } where

(26

i ∈ N | ∀i > 1

j ∈ (n = N i + 1)

⎧i − 1 Threshold = ⎨ ⎩ j −1 where

if

PCij > N C

if

PCij ≤ N C

(27)

i − 1 ≠ Un - Optimized Multihop j − 1 = Optimized Multihop N c = Infrastructure Costs

Pi (Node-i Power), Pj (Node-j Power), Dij (Node warranty &

The power consumption cost calculated as function of

Percentage of use) and E C (Node Energy cost) in equation (26), is compared with N C in (27) by the threshold point to determine if the multi -hop communication is optimized. The algorithm is prese ted in figure 5. 1

f u n c _ b lo c k 1

2

P a t h lo s s e x p o n e n t R e c e iv e S e n s it iv i ty E ne rgy C ost N o de C ost D u r a t io n

S e t I n p u t P a r a m e te r s

fu n c _ bl oc k 2 3

E n te r T x m i t d i s ta n c e

4

ho p k =

T x m it

d is t a n c e

/

k

6 5

h o p k > M a x d is ta n ce

Yes

ho p

k

d is ta n c e > lin k d i sta n c e

No 11 7

E v al h o p k E n e rg y c o n s u m p tio n ; h op k =

E v al h o p k S e p a ra tio n d is tan c e ;

N o t O p ti m iz e d

E v al h o p k T x m it p o w e r ;

f u n c _ b lo c k 3

ho pk

8

T x m it

po w e r


13

No

E v a l h o pk T x m it

9

p o w e r c o n s u m p tion c o st

1 00 m W

Y es

End

p o w e r c o n s u m p tio n c o st 12

hop

k

= O p tim iz ed

< N ode cos t

Y es 10

Figure 5: PMCA Alg orithm Flow Chart The PMCA consists of three functional blocks as depict d in Figure 5. A functional block (func_block1) that sets the appropriate values of the required input parameters, a functional block (func_block2) that determines the number of hops, TR separation distance of hops and transmit power level of hops, and a functional block ( func_block3) that determines the number of optimized hops.

ⓒGESTS-Jun.2010

8

Power-Optimization Method for the Implementation of IEEE …

In func_block1, the path loss exponent, receiver sensitivity and node adius are required as inputs. The path loss is representative of the environmental factors affecting he transmitted signal. The receiver sensitivity is an indicatio of the minimum signal required by the receiver to receive data at a specific data rate (throughput) given the path loss. For a given receiver sensitivity and the path loss, a specific tra smission range is achieved. The node radius is representative of this range. The receiver sensitivity and node radius are set for a single hop link in accordance with the maximum transmit power. In func_block2, the algorithm determines the number of hops for a specific transmission distance based on the single hop l parameter settings ( func_block1) . The numbers of hops are designated hopk , where hopk is expressed as: -

hop k ∈ k | f : k → Ν

(28)

The energy consumption, T-R separation distance between nodes and transmit power levels of nodes are then determined. In func_block3, the algorithm determines the power consumption cost ba d on equation (26) and utilizes the threshold in equat ion (2 7) to determine optimized and un-optimized number of hops V

P ERFORMANCE EVALUATION

In this section we evaluate the performance of the PMCA algorithm by conducting a comparative study. We compare a multihop communication system implemented conventionally and a multi-hop communication system implemented using the PMCA algorithm in order to validate the performance of the PMCA algorithm. All simulation tests were run in Matlab [12]. A. Simulation Scenario We created a scenario composed of infrastructure-based architecture with fixed Access Points (APs) for the simulations. The experimental setup is shown in Figure 6.

Figure 6: Conventional Multi-Hop Communication As shown in Figure 6, 20 APs are deployed in 1200 x 12 flat area. The distance between any two nodes is 200m. The transmit power of each node is set at a maximum of 100 W. For simplicity, the path indicated by a line from A to T on the graph is simulated. The path consists of 8 nodes operating under the DCF MAC layer scheme. This scenario represents the conventional multi -hop system. To effectively evaluate our PMCA algorithm, the perfor ance of the IEEE 802.11b was first evaluated. The conventional multi -hop system was then initialized in accordance with the IEEE 802.11b performance evaluation results. The performance of the PMCA a lgorithm was then compared with the perfo mance of the conventional system, which was based on the IEEE 802.11b performance evaluation results.

ⓒ GESTS-Jun.2010

GESTS Int’l Trans. Computer Science and Engr., Vol.60, No.1

9

B. Simulation Results This subsection provides results for the PMCA algorithm based on the previously discussed simulation scenario. 1. Simulation Parameter settings The simulation parameter settings for IEEE 802.11b and Conventional multi -hop system performance evaluation are described in table 1 below. The path loss exponent, node radius and node transmit power rameters are set in accordance with the scenario depicted in Figure 6. Table 1: Simulation Parameters

The IEEE 802.11b performance metrics models described in section III are implemented in MATLAB utilizing the parameters shown in table 1. 2. Propagation Model In order to validate the use of the log-distance model in our study, we conducted some practic measurements and compared them with the theoretical analysis of the model. We performed our measurements using off-the-shelf IEEE 802.11b hardware, the Linksys WAP54G wireless access point and a PDA. Th measurements were taken around the soccer field at Tshwane University of Technology, for distances varying from 1 to 100m. The experimental data and theoretical analysis results are shown in Figure 7. The results shown in Figure 7 indic that the log-distance model can be successfully used to yield accurate results in our study. Firstly, the value found for the path loss exponent (n = 2.8) agree with the values published in [10] because the area we took measurements within is r resentative of an urban area or factory side area with obstructions. Secondly, our results produced an acceptable standard viation ( s ) of 3.0797dB. 3. IEEE 802.11b Performance evaluation The performance of a communication system in a wireless environment is dependent on the wireless channel. The channel conditions determine the quality and the reliability of the transfer of information from the sender to the receiver. In IEEE 802.11b, throughput is utilized as a performance measu e of a communication system and is dependent on the transmission rate and the wireless channel conditions. The sender’s MAC layer dynamically chooses the transmission rate for each transmitted frame based on the wireless channel condit ons. To evaluate channel conditions, the RSS, SNR and BER are analysed. The RSS depends on the transmitted power and the path loss. The SNR is dependent on the RSS. The BER for a given modulation scheme that determines the transmission rate is calculated from the SNR at the receiver. Throughput is a function of the ratio of the MSDU length and the average transmission time for a single frame (Tfra me ). The modelling of the IEEE 802.11b throughput was discussed in section III.

ⓒGESTS-Jun.2010

10

Power-Optimization Method for the Implementation of IEEE …

Figure 7: Comparison of Measured Path loss and Theoretical Path loss

Figure 8 shows the IEEE 802.11b RSS performance as a f nction of distance. The path loss exponent and the node transmit power are set as indicated in table 1.

Figure 8: Simulated IEEE 802.11b RSS The result shown in Figure 8 indicates that the RSS de ys exponentially with distance. Thus, the node receiver sensitivity will also vary with distance. From Figure 8, the node receiver sensitivity for IEEE 802.11b given the channel conditions depicted in table 1 is specified in table 2. Table 2: IEEE 802.11b Receiver Sensitivity Specification Receiver Sensitivity

Distance

-88dBm -91dBm -96dBm -98dBm

200m 250m 350m 400m

The SNR is determined from equation (9) shown in secti n III. From equation (9) it is shown that the SNR is directly proportional to RSS since we assumed an AWGN channel. Thus, it also decays exponentially with distance as shown in Figure

9.

ⓒGESTS-Jun.2010

GESTS Int’l Trans. Computer Science and Engr., Vol.60, No.1

11

Figure 9: Simulated IEEE 802.11b SNR The SNR for the node receiver sensitivity specification depicted in table 2 is shown in table 3. Table 3: IEEE 802.11b SNR Performance

SNR

Receiver Sensitivity

Node Radius

14dB 11dB

-88dBm -91dBm -96dBm -98dBm

200m 250m 350m 400m

7dB 5dB

The results shown in table 3 indicate that shorter links are more reliable and stable since they are characterized by higher values of SNR. For instance, an SNR of 5dB is achieved when covering a 400m distance with a single hop transm sion. However, if two or three hops are used the SNR can be ncreased up to 14dB, resulting in link quality improvements. Thus, shorter links are desirable as they improve the reliability of link which is dependent on the channel conditions (SNR). The BER for a given modulation scheme is determined from the received SNR shown in Figure 9. The performance results are

shown in Figure 10.

Figure 10: Simulated BER for IEEE 802.11b

ⓒGESTS-Jun.2010

12

Power-Optimization Method for the Implementation of IEEE …

It is shown in Figure 10 that the transmission rate is dependent on the channel conditions (SNR). As the channel conditions degrades, a lower rate is chosen by the MAC layer to reduce the BER. If the SNR drops from 14dB to 5dB, the transmission rate is reduced from 11Mbps to 1Mbps to keep the BER at a minimum of 10 -7 . Otherwise, if the same transmission rate is - 1 .2 utilized, then the BER will be increased to approximately 10 and this will result in packets being lost. The transm ssion rate selection given the channel conditions (SNR) indicated in table 3 is shown in table 4. Table 4: Transmission rate selection based on channel conditions Modulation Technique CCK CCK

Transmission

SNR

Distance

14dB 11dB

200m 250m 350m 400m

Rate 11Mbps 5.5Mbps 2Mbps 1Mbps

QPSK BPSK The results shown in table 4 suggests that shorter lin

7dB 5dB

sults in improvements in transmission rates. A transmission rate of

1Mbps is achieved with a single hop transmission (400m . However, when utilizing multi-hop transmission, the transmission rate can be increased up to 11Mbps. Thus, shorter link are desirable for high data rate transmission. The throughput performance for different transmission rates based on channel conditions is shown in Figure 11.

Figure 11: Simulated IEEE 802.11b Throughput It is shown in Figure 11, that the actual expected throughput for each transmission rate is drastically reduced, particularly for the 11Mbps rate. The popular IEEE 802.11 WLAN is known to achieve relatively small throughput performance comp ed to the underlying PHY layer transmission rate [17]. This s caused mainly by the MAC/PHY layer overheads such as MAC header, PLCP preamble and header, ACK transmission and some IFS. Since the overheads are added to each frame transmission, the throughput degradation is relatively when the small -size frames are transmitted. The IEEE 802.11b throughput performance based on the results shown in table 2 -4 is shown in table 5. Table 5: The IEEE 802.11b Throughput Performance

Data Rate

SNR

11Mbps 5.5Mbps 2Mbps 1Mbps

14dB 11dB

7dB 5dB

Receiver Sensitivity -88dBm -91dBm -96dBm -98dBm

Max Throughput 6.15Mbps 3.85Mbps 1.7Mbps

950Kbps

Distance 200m 250m 350m 400m

Table 5 shows the maximum achievable IEEE 802.11b throughp t under the specified channel conditions. It is shown in table 8.5 that shorter links result in increased transmission rates and consequently improved throughput performance. In essence, t he

ⓒGESTS-Jun.2010

GESTS Int’l Trans. Computer Science and Engr., Vol.60, No.1

13

results shown in table 2-5 indicates that the performance of a multi -hop system can be improved by using several shorter li ks instead of long distance links to implement the system

4.The PMCA Algorithm Evaluation The PMCA based multi-hop system is simulated based on the performance evaluation of the IEEE 802.11b d cussed in the previous subsection. The objective is to compare the t ansmit power consumption efficiency of the conventional multi -hop system and the PMCA based system. It should be recalle that the distinguishing factor for the two systems is that in conventional multi -hop system, transmit power levels of nodes are normally randomly set to maximum levels that link the nodes together to extend the network coverage area and in PMCA based systems the Power-efficient Multi -hop Communication Algorithm (PMCA) is used to set appropriate transmit power levels of nodes to optimize transmit power consumption. In the scenario depicted in Figure 6, a throughput of 6.15Mbps is achieved in a single hop transmission as shown in table 5. To extend the range to 1400m with the same throughput erformance, 7 hops are utilized as shown in Figure 6. This scenario represents the conventional multi -hop system. Figure 12 indicates that transmit power consumption is significantly reduced when the PMCA algorithm is utilized implement the multi -hop systems. In the PMCA based system, the distance between each hop is reduced to optimise nodes transmit power. However, the reduction of distance res lts in the reduction of the coverage area. Thus, additional nodes are added to cover the same range as in the conventional s tem. This approach optimises power consumption in multi -hop communication. As discussed in section IV, there is a specific number of hops that is optimized in terms of power consumption for any multihop communication system. The PMCA algorithm is utiliz d to establish these hops. In Figure 12, it is shown that 8 hops are optimized to cover the same area covered by the conven al system. The comparison results are shown in table 6.

Figure 12: PMCA Based Multi -Hop System vs. Conventional Multi-Hop System The results shown in table 6 indicate that the convent onal system consumes 800mW of power while the PMCA based system consumes 657mW of power. Thus, 18% of transmit power consumption is saved in the PMCA ased system. Table 6: Comparison of PMCA based system and Conventional sys em No of

Node Tx power

Node Radius

100mW 73mW

200m 175m

Nodes 8 9

Additional nodes increases infrastructure costs which add to overall system costs (power consumption and infrastructure). In Figure 13, it is shown that overall system costs of the PMCA based system are considera bly lower when compared to the conventional system. This is due to the fact that power consumption is gnificantly reduced in the PMCA based system.

ⓒ GESTS-Jun.2010

14

Power-Optimization Method for the Implementation of IEEE …

Figure 13: System cost vs. Hops C. Revisal by Measurements We verified the achieved simulation results by conducting some practical measurements. We further evaluated the effect of optimized transmit power levels on energy consumption interference. For these purposes the measurements were conducted using Linksys wireless broadband routers (WRT54GL). A Freifunk firmware version v.1.2.5 was used to run the router and change parameter settings. For benchmarking PUTTY version v.0.58 was used. PUTTY is a free Secure Shell (SSH), Telnet and Rlogin network protocol software. Throughput was used as the performance metric measurement. A default value of 8192 bytes was used for the length of the buffer. 1. Transmit Power The measurements for transmit power consumption were taken at a location within the vicinity of Tshwane University of Technology in Pretoria. Two scenarios were evaluated in an outdoor environment. In the first scenario a distance of 500m was covered using two nodes. In the second scenario the sa e distance is covered using three nodes. The results are shown in table 7. Table 7: Conventional approach vs. Proposed approach

Nodes

Node Tx

2 3

64mW 10mW

No of

Transmission

Throughput

Rate

Node

Radius

Power

202Kbps 250Kbps

194Kbps 242Kbps

500m 250m

The results in table 7 shows that in a conventional approach, 128mW of transmit power is consumed since each node utilized a maximum transmit power of 64mW. In the proposed appr ach, 30mW of transmit power is consumed since each node utilized an optimum transmit power level of 10mW. Thus, approximately 77% of transmit power is saved for the same throughput performance. The throughput performance for the two scenarios sli htly differs due to the nature of the wireless channel. As a result, average values were taken.

2. Energy Consumption To evaluate the effect of transmit power on energy consumption, the energy consumption of the device needs to be known. Since we do not have any access to the energy consumption characteristics of any IEEE 802.11b compliant products currently available in the market, the current drawn when the de ce was not transmitting (idle mode) was measured and the current drawn when the device was transmitting (transmission m de) was measured. The results are as shown in table 8. The energy consumption for both modes was calculated u ing the following standard formulae

PI _ mod e = V × I I _ mod e

ⓒGESTS-Jun.2010

(29)

GESTS Int’l Trans. Computer Science and Engr., Vol.60, No.1

15

Where PI_m od e is the idle mode power, V is the voltage supply and II_ mo de is the idle mode current. Similarly

PT _ mod e = V × I T _ mod e

(30)

Where PT_m od e is the transmission mode power, V is the voltage supply and I T_ mo d e is the transmission mode current. From Equation (30)

E I _ mod e = PI _ mod e × t

(31)

Where EI_m od e is the idle mode energy consumption and t is time in seconds. From Equation (31)

E T _ mod e = PT _ mod e × t

(32)

Where ET_m od e is the transmission mode energy consumption and t is time in seconds. Table 8: Power Characteristics of the Linksys WRT54GL Idle Mode

Current 245mA

Transmission Mode Current 320mA

Max Tx P ower 64mW

Power Supply 12V

The energy consumption for the two modes is shown in t le 9. The results are for nodes transmitting continuously for 12 seconds. Table 9: Idle Mode and Transmission Mode Energy Consumption Idle Mode Energy Consumption

Trans mission Mode Energy Consumption

2.94J

0.9J

The results shown in table 9 indicates that the RF energy consumption (transmission mode) is approximately 25% of the total energy consumed when a maximum level of transmit power used. For a short range coverage, fewer hops are used and for a long range coverage, several hops are used. Thus, in t rms of energy consumption, the proposed algorithm is more applicable to long range coverage. To evaluate energy consumption for both short range coverage and long range coverage, Equation (34) below is used.

E = N (Eidle + ERF )

(33 )

Where N is the number of nodes, Eidle is the idle mode energy consumption and ER F is the RF energy consumption.

ⓒGESTS-Jun.2010

16

Power-Optimization Method for the Implementation of IEEE …

Figure14: Energy consumption analysis for short range

rage

In Figure 8.8, it is shown that 300% of energy is consumed when 3 nodes with maximum node transmit power levels are used. An extra node is added and the distance between nodes is halved when utilizing the proposed approach. Thus, node transmit power is reduced by 77% as discussed in sub-section 1. Energy consumption is increased to 323% as n in Figure 8.8 resulting in energy consumption being increased by 7.3% in the proposed approach.

Figure15: Energy consumption analysis for long range c

rage

In Figure 15, 800% of energy is consumed when 8 nodes with maximum node transmit power levels are used. However, when an extra node is added, energy consumption is reduced to 727%. Resulting in energy consumption savings of 9.1%. In es nce the comparison of results shown in Figure 14 and Figur 15 indicates that the proposed approach has energy consumption benefits in a conventional system with a high density of nodes. 3. Interference To evaluate the effect of transmit power on interference, measurements were conducted at the Meraka Institute in CSIR,

Pretoria. The setup was similar to sub-section 1. However, the measurements were conducted in an indoor interference prone environment with a maximum single hop distance of 100m. The

sults are shown in table 10.

Table 10: Transmit power vs. Interference

No of Nodes 2 3

Node Tx Power 64mW 8mW

Packet Loss 75% 0%

Node Radius 100m 50m

The results shown in table 10 indicate that in an interference prone environment, optimized transmit power levels can even revive an apparently “dead” link. In a single hop link 75% of packets were lost. In a multi-hop link, the transmit power level of

ⓒGESTS-Jun.2010

GESTS Int’l Trans. Computer Science and Engr., Vol.60, No.1

17

nodes was significantly reduced and the link was revived. Thus, the use of power -efficient multi-hop communication systems not only saves power and energy, but can also be used s an interference minimization mechanism for the congested 2.4GHz ISM spectrum. From the achieved results, we can deduce that the PMCA algorithm provides the following benefits: 1.

It offers a mechanism that enables range extension of WLANs with optimized transmit power consumption. The use of optimized transmit power levels provides significant energy consumption savings as shown in the achieved results. Furthermore, optimized transmit power levels contribut significantly in the reduction or minimization of interference

within the congested IEEE 802.11 2.4GHz ISM spectrum. 2.

It provides a solid framework for the development of IEEE 802.11 RF planning tools such as ATOLL (from FORSK) used for the GSM technology. Such planning tools are r e or non-existent in IEEE 802.11. Thus, the PMCA algorithm is provided as a simple RF planning mechanism that facilitates the analysis, planning and implementation of op imized IEEE 802.11 multi-hop networks.

VI CONCLUSION AND F UTURE WORK In this paper, we presented an algorithm that ensures plementation of IEEE 802.11b multi -hop networks with optimized power consumption. Simulation results have shown that power c sumption is significantly optimized in PMCA based multihop systems as compared to conventional multi -hop system. We have also shown through the conducted practical experiments that power optimization can lead to energy savings in multi-hop networks with a higher density of nodes (long range coverage). Furthermore, it was shown that lower levels of transmit power result in lower probabilities of interference within the interference prone ISM band such that “apparently dead” links can even be revived. Thus, the algorithm delivers a significant framework for the implementation of IEEE 802.11 multi -hop communication networks. The IEEE 802.11 WLAN’s are envisioned as providers of permeate and use seamless communication in future wireless broadband technologies, hence, improving their performance is a ogical, significant and a future save concept. The work on this paper does not take into consideration the routing and signa propagation issues prevalent in multi-hop networks. Thus, further investigations needs to be carried out in relation to the above mentioned issues.

ACKNOWLEDGEMENTS The authors would like to thank Telkom Centre of Excel nce Program and F’SATIE at Tshwane University of Technology for making this research possible.

REFEREN CES [1] [2] [3]

[4] [5] [6]

Hao Zhu, Guo ho ng Cao , “O n Impro ving the P erformance o f IEEE 80 2.11 with Relay-Enabled PC F,” Mob ile N etwo rk s and Applications, vol. 9 , no. 4, p p. 42 3 - 43 4, 2 00 4. Chan, A. H., "Benefits and Requirements of Interworking between WirelessLAN and 3G Wireless," P roceedings of Southern African Telecommunication Network s & Ap plications Co nference (S ATNAC20 04 ), So uth Western C ap e, So uth Africa, 6 - 8 S ep temb er 20 04 , Vol 1, pp. 14 9 - 15 3 R. Pabst, B.H. Walk e, D. C . S chultz, P. Herhold , H. Yaniko mero glu, S . Mukherjee, H. Viswanathan, M. Lott, W. Zirwas, M. Do hler, H. Aghvami, D. D. F alconer, and G. P. F ettweis, “Relay- based deployment concepts for wireless and mobile broadband radio”, IEEE Communications Magazine, September 2 00 4 R. Ramanathan and R. Rosales-Hain, “ Top ology Co ntrol of Multiho p Wireless Network s using Transmit Po wer Ad justment,” Proceedings o f the IEEE Conference on C omputer Communications (INFOCOM), pp. 4 04 -4 13 , Tel Aviv, Israel, March 2 00 0 Michael Fainberg, “A Performance Analysis of the IEEE 802.11b Local Area work in the Presence of Bluetooth PersonalArea Network”, Master Thesis (Telecommunication N etworks), Polytechnic University – Brooklyn, June 2001 Jonas Linnell, Hans Torstensson, “Radio Resource Management in IEEE 802.11b WLAN : Aspects of System Planning”, Master Thesis, Chalmers University of

Technology, Götenborg, Feb 2004. [7] [8] [9] [1 0]

[1 1] [12] [1 3]

Javier del Prado P avon, S unghyun C hoi, “Link Adaptatio Strategy for IEEE 802.11 WLAN via Received S ignal Strength M easurement”, Proceedings of ICC’0 3, Vol. 2, pp . 1 108 - 111 3, Anchorage, Alaska, US A, May 20 03 Liljana Gravrilovska, Vladimir Atanasovski, “Influence of Packet Length on IEEE 802.11b Throughput Performance in Noisy C hannels”, Pro ceed ings of 1 st Intern ation al MAGNET Work sh op D6.2.5 , S hangai, China, 1 1 - 12 N ovember 20 04 , ISBN 87 -9 16 96 -3 6 - 4. Qian Wu, Carey Williamso n, “Dynamic Channel Rate Assignment for Multi-Rad io WLAN’s” Proceedings of the IAS TED co nference o n wireless netwo rks and emerging technologies (WNET), pp. 149 -1 54 , Banff, Alberta, C anad a, July 2 00 5. T.S . Rap papo rt, Wireless Co mmunicatio ns: P rincip les and Practice, S econd Edition, (Upper Saddle River: P rentice Hall P TR, 2 002 ). http ://www.cisco .co m/en/US/pro ducts/hw/wireless/ps45 70 p rod ucts_ data_sheet091 86 a0 08 00 f9 ea7 .html. Matlab, language for technical computing, information available at http://www.mathworks.co m A technical rep ort C U-C S - 93 4 - 02 , o n the imp lementatio n o f transmit p ower contro l in

ⓒGESTS-Jun.2010

18

Power-Optimization Method for the Implementation of IEEE …

802.11b wireless networks, http://mantis.cs.colorado.edu/med ia/CU_CS _9 34 _0 2 _ Transmit_ Po wer.pdf [14] Bright C hu, “Improving IEEE Performance with Power Con and Distance- b ased C ontentio n Wind ow S election”, Master Thesis, University of Illino is, Urb ana, 20 04 . th [1 5] E.- S Jung and N.H Vaidya, “ A Po wer Contro l Mac P ro to col fo r Ad Ho c Network s,” In pro ceed ings of th e 8 ann ua l A CM/IE EE Intern ation al Co nference o n Mo bile Co mpu tin g an d Networkin g (MOBICOM), Septemb er 20 02, pp . 3 6 - 47 . [1 6] F . Ab desslem, M. Dias De Amorim, K. K ab assanov and S. Fdid a, “ On the feasibility o f p ower co ntrol in current IEEE 80 2.11 d evices,” In proceedings of secon d IEEE Percom Wo rk shop on P erva siv e Wireless Netwo rk ing (PW N’06). Pisa (Italy), March 200 6. [1 7] Y. Kim, S. Cho i, K. Jang, H. Hwang, “Throughput enhancement of IEEE 802.11 WLAN via F rame Aggregation,” In pro ceed in gs of Vehicu lar Tech nolo gy Conference (VT C’04 -Fall), September 2 00 4, pp . 3 03 0 - 30 34. ISBN : 0 -7 80 3 - 85 21 -7.

ⓒGESTS-Jun.2010