Cross-Layer Cooperative Triple Busy Tone ... - Semantic Scholar

Cross-layer Cooperative Triple Busy Tone Multiple Access for Wireless Networks Hangguan Shan† , Ping Wang§ , Weihua Zhuang∗ , and Zongxin Wang†

† Communication Science and Engineering Department, Fudan University, Shanghai, P. R. China, 200433 Email: [email protected] [email protected] § School of Computer Engineering, Nanyang Technological University, Singapore, 639798 Email: [email protected] ∗ Department of Electrical and Computer Engineering, University of Waterloo, Waterloo, Ontario, Canada, N2L 3G1 Email: [email protected]

Abstract—In this paper, with the cross-layer design principle, a novel cooperative triple busy tone multiple access (CTBTMA) scheme is proposed for wireless networks to achieve cooperative diversity gain. A utility-based algorithm is presented to determine the capability of a node in helping other nodes’ transmissions. With the use of three busy-tone channels, not only collisions can be avoided, but also an optimal helper can be determined without disturbing existing transmissions. Simulation results demonstrate that the proposed scheme can effectively increase the throughput in a low SNR environment, as compared with IEEE 802.11a single-hop transmissions. On the other hand, transmit power can be greatly reduced in the proposed scheme in order to achieve the same throughput as in the single-hop transmissions.

I. I NTRODUCTION Cooperative diversity is emerging as a powerful technique for improving the reliability and throughput of wireless networks [1-4]. The basic idea is that nodes in a wireless network share their information and transmit cooperatively as a virtual antenna array, thus providing diversity without the requirement of additional antennas at each node. When cooperative diversity is used in wireless networks, a cooperation-based medium access control (MAC) scheme needs to be carefully designed. First, in the cooperative communication scenario, a helper (which is a relay node that helps the source node to deliver its packets to the destination node) not only receives packets from the source but also transmits the packets to the destination. The transmissions from neighbours of the helper should be carefully scheduled to avoid collisions. Otherwise, the cooperation gain will be reduced. Second, there may exist a number of helpers that can potentially improve the transmission quality (e.g., resulting in higher throughput and lower bit error rate) from a source to a destination. Without a central controller, how to find the optimal one(s) effectively and efficiently is vital to a practical MAC protocol. To the best of our knowledge, there is no cooperation-based MAC addressing these problems so far. In [5], a cooperative-diversity slotted ALOHA (CDSA) considering routing cost is presented for ad hoc networks. When direct transmission fails, a potential relay that receives the packet from the source contends to be the helper if its routing cost is less than that of retransmission from the source node. In [6] and [7], to mitigate the throughput This research work was supported by a fund form China Scholarship Council (CSC) and by a research grant from the Natural Science and Engineering Research Council (NSERC) of Canada.

bottleneck caused by low date rate nodes, the authors propose cooperative and relay-enabled MAC schemes, respectively, based on the IEEE 802.11 DCF. In both schemes, high rate nodes are allowed to help low data rate nodes through twohop transmissions, but the selection of a high rate helper may not be optimal as it is based on observation of historical transmissions. Furthermore, the exchanges of observation or waiting for an unresponsive helping request may result in inefficiency. In this paper, to solve the above two problems, inspired by the dual busy tone multiple access (DBTMA) [8, 9], we propose an efficient cooperative triple busy tone multiple access (CTBTMA) scheme for wireless networks, where adaptive modulation and coding (AMC) and transmission type (i.e., cooperative transmission or two-hop transmission) are crosslayer designed. In DBTMA, request-to-send (RTS) and DATA frames are protected by dual busy tones, i.e., the transmit busy tone BTt and the receive busy tone BTr, respectively. We adopt these two busy tones in our scheme. In addition, a helper busy-tone channel BTh is added to avoid collisions at the helper and to fulfill the optimal helper contention or selection1 . The criterion under consideration for an optimal helper is to maximize instantaneous throughput from the source to the destination. The remainder of this paper is organized as follows. In Section II, the proposed CTBTMA protocol is presented. In Section III, a cross-layer design approach is given to select a helper that maximizes the instantaneous throughput from a source to a destination. Numerical results are given in Section IV, followed by the conclusion remark in Section V. II. P ROTOCOL D ESCRIPTION Taking the network shown in Fig. 1 as an example, we explain the operation procedure of the proposed CTBTMA protocol. In this example, node A wants to send data to node B, while node G has already been sending data to node F. In the figure, a solid line between any two nodes indicates that they can hear each other, and an arrow denotes a transmission. When node A has data to send, it first senses the three busy-tone signals, BTt, BTr, and BTh. If no busy-tone signal 1 As a multi-helper scenario requires accurate time synchronization among different helpers, here we consider a single-helper case.

978-1-4244-2324-8/08/$25.00 © 2008 IEEE. This full text paper was peer reviewed at the direction of IEEE Communications Society subject matter experts for publication in the IEEE "GLOBECOM" 2008 proceedings.

K C H

A

B

I

J

E F

Fig. 1.

D

G

An example network to illustrate cooperation

is detected, which means that none of the nodes in its transmission area (i.e., nodes B, C, D, E, and H) is in the packet reception state, node A turns on its BTt signal, sends an RTS packet to node B, sets up a timer TS1 , and waits for a response from node B. The timer is given by TS1 = TRT S + τ + TCT S + τ

(1)

where TRT S and TCT S are the transmission times of RTS and clear-to-send (CTS) packets, respectively, and τ is the propagation delay. If no CTS packet from node B is received by node A before the timer TS1 expires, node A will turn off its BTt signal and give up the transmission. When node B receives the RTS packet, if neither BTr nor BTh signal is detected, it first turns on its BTr signal as an indication of a successful reservation, and then sends back the CTS packet including its estimated signal to noise ratio (SNR), and sets up a timer TD1 as TD1 = TCT S + τ + Td + τ + TDAT A + τ

(2)

where Td is the busy-tone detection delay which depends on the communication hardware and might not be negligible, TDAT A is the transmission time of a data packet that can be calculated based on the date transmission rate and packet length. The timer TD1 is used to account the time in which node B should receive the data packet from node A if no helper exists. If no data packet is received in TD1 , node B turns off its BTr busy tone. Node A monitors the BTr signal and waits for the CTS packet from node B. Once the CTS packet is received, node A sets up a new timer TS2 = Td + τ , and waits for response from potential helpers. If there is no BTh signal detected before the timer TS2 expires, which means that no helper has the ability to improve the instantaneous throughput of transmission from node A to node B, node A turns off its BTt signal and starts to send the data packet to node B with the transmission parameters (e.g., modulation mode and coding rate) chosen according to the SNR information included in the CTS packet. Otherwise, node A keeps transmitting its BTt signal until it receives the ready-to-help (RTH) packet from an optimal helper. Note that different from DBTMA, where an existing receiver (e.g., node F in Fig. 1) turns on its BTr busy tone during the data packet reception, in our scheme it is the potential receiver (e.g., node B) which just received an RTS packet that turns on its BTr busy tone to protect the reception of the RTH packet from the optimal helper. For an existing receiver (e.g.,

node F) during its packet reception, it will turn on the BTh busy tone (rather than the BTr busy tone) to protect its data packet reception. If all the receivers (e.g., nodes F and B) use a BTr busy tone to protect their receptions, a helper cannot distinguish whether a BTr busy tone is sent by its own receiver or by any other existing receiver. Next, we discuss how to choose an optimal helper without interfering with other existing transmissions. Assume that nodes C, D and E all have the ability to improve the instantaneous throughput of the transmission from node A to node B. The ability of a node to cooperate a transmission in CTBTMA is measured by a utility value (to be explained in Section III). The larger the utility value, the better the ability. Any node that receives both RTS and CTS packets and does not sense a BTh signal calculates its utility value according to the algorithm in Section III. In the example of Fig. 1, because of the BTh signal from the existing receive node F, node E will not contend to be a helper in order not to corrupt the reception of node F. On the other hand, without sensing a BTh busy tone, nodes C and D will contend to be a helper. They send out their busy tones in the BTh channel. The length (i.e., time duration) of the BTh busy-tone signal is a function of the utility value. The larger the utility value, the longer the BTh busy-tone transmission. After exhausting its BTh busytone signal, a potential helper (say node C) senses the BTh channel. If the BTh channel is busy, which means that at least one other node is transmitting a longer BTh busy-tone signal (implying that a better helper exists), node C will give up. If the BTh channel is idle, which means that node C is the optimal helper, it will send an RTH packet including the optimized transmit parameters (transmission type, modulation and coding scheme) to node A and node B. At the same time, node C continues to send its BTh busy-tone signal to protect its packet reception during cooperation. Upon receiving the RTH packet, the destination (e.g., node B) switches its busytone signal from the BTr channel to the BTh channel, and the source (e.g., node A) turns off its BTt busy tone and starts to send a data packet. If a destination does not detect a BTh busy tone within τ + TS2 after sending a CTS packet (which means that no helper exists), it switches its busy tone signal from the BTr to the BTh channel. After receiving (sending) the RTH packet, node B (node C) initiates a timer TD2 (TH ) given by SD if cooperative tx TDAT A + τ TD2 = HD SH TDAT A + τ + TDAT A + τ if two-hop tx (3) SH TH = TDAT + τ (4) A SD where TDAT A is the transmission time of a data packet SH HD through the source-destination channel, TDAT A and TDAT A are the transmission times of a data packet through the sourcehelper channel and helper-destination channel, respectively. In cooperative transmission, a destination receives the data signals from both the source and the helper. In two-hop transmission, a destination receives the data signal only from the helper. Once the destination receives the data packet, it


sends an acknowledgement (ACK) packet to the source node. If no data packet is received by node C (node B) after timer TH (TD2 ) expires, node C (node B) turns off its busy-tone signal. Fig. 2 illustrates the operation procedure of CTBTMA in the example network shown in Fig. 1, where cooperative transmission is adopted by node A. III. C ROSS - LAYER D ESIGN In this section, we introduce the algorithm to determine the ability that a potential helper has to help a transmission node pair. A potential helper here is a node hearing both RTS and CTS packets from the source and the destination, respectively, and not detecting any BTh busy-tone signal. It gets the SNR information of the source-helper channel, sourcedestination channel, and helper-destination channel by either measuring the SNR of the received RTS/CTS packet or directly from the content of CTS packet, denoted by γSH , γSD , and γHD , respectively. Note that a destination estimates the sourcedestination channel, and feedbacks the SNR in its CTS packet. Consider that the channel is reciprocal, the SNR of the helperdestination channel is estimated via the destination-helper channel. We define the utility function as the instantaneous throughput of a transmission, given by W · TP + TO [1 − Pe (γSH , γHD , γSD , W, r)]

U (H, r|γSH , γHD , γSD , W ) =

(5)

where W is the data payload in bits, TP is the total payload’s transmission time, TO is the total overhead transmission time including the propagation delay, RTH packet, and BTh busy tone (we ignore the common items, i.e., TRT S and TCT S ), r is the effective payload’s transmission rate which depends on the transmission type, modulation and coding scheme, Pe (γSH , γHD , γSD , W, r) is the packet error rate at the destination, as a function of γSH , γHD , γSD , W , and r. Next we derive TP , TO and Pe (γSH , γHD , γSD , W, r). Consider packet error rate for convolutional coding as it has good error correct capability and is adopted in several standards such as [10-12], and suppose that hard-decision Viterbi decoding is used at the receiver. For a single-hop transmission, over the source-destination channel, we have T1,P = W/R1,SD T1,O = TS2 + LO /rb + τ P1,e (γSH , γHD , γSD , W, R1,SD ) = Peu (γSD , W, R1,SD )

(6)

where R1,SD is the effective payload transmission rate, rb is a basic rate for transmitting the overhead of length LO , and Peu (γSD , W, R1,SD ) is the upbound decoding error at given SNR, payload length, and transmission rate, given by Peu (γ, W, r) = 1 − (1 − Pu (γ))W

(7)

with Pu (γ) being the union bound of the first-event error probability. Note that transmission rate r depends on modulation

and coding. Given the modulation and coding scheme (i.e., given r), Pu (γ) can be written as ∞ ad · Pd (γ) (8) Pu (γ) = d=df ree

with df ree being the free distance of the convolutional code and ad is the total number of error events of weight d, which can be obtained from [13]. For hard-decision decoding, Pd (γ) is given by [14] ⎧ d d k ⎪ d−k ⎪ if d is odd ⎪ k ρ (γ)(1 − ρ(γ)) ⎪ ⎪ k=(d+1)/2 ⎨

1 d d/2 (γ)(1 − ρ(γ))d−k + Pd (γ) = 2 d/2 ρ ⎪ ⎪ d d k ⎪ if d is even ⎪ d−k ⎪ ⎩ k ρ (γ)(1 − ρ(γ)) k=d/2+1

(9) where ρ(γ) is the bit error probability for a given SNR γ under AWGN. For two-hop communications, a data packet is relayed by the helper, denoted by H, through the source-helper-destination channel. We have T2,P = W/rSH + W/rHD T2,O = TBT h (H) + τ + TRT H + τ + 2 · (LO /rb + τ ) (10) where rSH and rHD are the payload’s transmission rates in the source-helper channel and helper-destination channel, respectively, TBT h (H) is the time duration of H’s BTh busy tone. The effective payload transmission rate for the two-hop communication is given by R2,SD = rSH · rHD /(rSH + rHD ).

(11)

The packet error rate for two-hop communications is then [1 −

P2,e (γSH , γHD , γSD , W, R2,SD ) = 1− u Pe (γSH , W, rSH )] · [1 − Peu (γHD , W, rHD )].

(12)

For cooperative communication, since different protocols affect the packet error, we consider two practical protocols [2]: CRC-based decode-and-forward (CDF) and amplify-andforward (AF). In CDF, a helper forwards the source data packet to the destination only if it decodes the data packet correctly. In AF, a helper simply amplifies the received data signal from the source and forwards it to the destination. The destination combines the received packets from the sourcedestination channel and helper-destination channel. A detailed description on cooperative communication can be found in [2]. For CDF, we have TCDF,P = 2 · W/rs TCDF,O = TBT h (H) + τ + TRT H + τ + 2 · (LO /rb + τ )· (1 − Peu (γSH , W, rs )) + (LO /rb + τ ) · Peu (γSH , W, rs ) PCDF,e (γSH , γHD , γSD , W, RCDF,SD ) = Peu (γSH + γSD , W, rs ) · [1 − Peu (γSH , W, rs )]+ Peu (γSD , W, rs ) · Peu (γSH , W, rs ) where rs is the payload’s transmission rate by the source. Here two cases that the helper can or can not decode the data


TS2 TS1 BTt

Node A

RTS

CTS

RTH

DATA

BTh BTr

Node B

RTS

Node C

RTS

CTS

Node D

RTS

CTS

Node E

RTS

CTS

CTS

RTH

DATA

DATA

ACK

BTh BTh

RTH

DATA

DATA

BTh

BTh

Node F

ACK

Node G

DATA

Fig. 2.

The time diagram of the CTBTMA protocol in the network shown in Fig. 1

packet from the source are considered in the calculation of TCDF,O and TCDF,e ; and the SNR values of the two cases at the destination are (γSH + γSD ) and γSD , respectively. For AF, we have TAF,P = 2 · W/rs

Δt ≥ Td . Note that Δt should be large enough to avoid an error in BTh judgement due to the propagation delay between two helpers. Finally, based on (5), any potential helper computes U (H, r|γSH , γHD , γSD , W ). max

TAF,O = TBT h (H) + τ + TRT H + τ + 2 · (LO /rb + τ ) PAF,e (γSH , γHD , γSD , W, RAF,SD ) = Peu (f (γSH , γHD , γSD ), W, rs )

If the maximal utility value of a potential helper is achieved by the two-hop or cooperative communication rather than the single-hop communication, it sends out a BTh busy tone to contend to be a helper, the length of which is given by (15) .

where f (γSH , γHD , γSD ) = γSD +γSH γHD /(γSH +γHD +1) is the SNR of maximal ratio combined (MRC) signal at the destination using AF. For the orthogonal transmission in two slots, the effective payload rates for the CDF and AF are half of the payload’s transmission rate RCDF,SD = RAF,SD = rs /2.

(13)

From the definition of the utility function (5), it is clear that any utility value satisfies U (H, r|γSH , γHD , γSD , W ) < r.

(14)

As the optional effective payload transmission rates are always finite in practice, denote them by R1 , R2 , . . . , RK , where K is the number of possible rates. One possible mapping function between the utility value and the length of BTh busy tone TBT h (H) is linear mapping as TBT h (H) = TS2 +

U (H, r|γSH , γHD , γSD , W ) − Ri−1 i−1+ 2· Δt Ri − Ri−1 if U (H, r|γSH , γHD , γSD , W ) ∈ [Ri−1 , Ri ) (15) where x is the ceiling function, Δt is the minimal differentiable time between two busy-tone signals and should satisfy

{modulation, coding rate}

IV. N UMERICAL R ESULTS To evaluate the performance of the proposed scheme, we compare the throughput performance of the proposed scheme with that of the IEEE 802.11a [11] in a single-hop transmission scenario where a transmitter (the source) and a receiver (the destination) are stationary and separated by a distance dSD = 200 meters. In every transmission, four potential helpers are randomly generated along the line that connects the source and the destination. The channel of any transmission pairs is modeled by joint log-distance (with exponent α=4 in the simulation) and Rayleigh fading. In performance comparison, the parameters of the proposed scheme are set in a conservative way as compared with IEEE 802.11a parameters, listed in Table I. According to IEEE 802.11a, the data rates vary from 6 to 54Mbps in the simulation, as shown in Table II. Fig. 3 shows the throughput gain of the proposed CTBTMA protocol over the IEEE 802.11a single-hop transmission. The average SNR is defined as SN R = Pt /d4SD N0 , where Pt and N0 are the transmit power and the received noise power, respectively. It can be seen that at a low SNR, the proposed scheme has a higher throughput than the IEEE 802.11a singlehop transmission. When the channel between the source and destination is poor, a helper can effectively increase the


TABLE I S IMULATION PARAMETERS Value 9μs 16 μ s 34 μ s 16 μ s 4μs 4μs 16 μ s 9μs 9μs

Notes Slot time SIFS time DIFS time = tSIFS+2 · tSlot PLCP preamble duration PLCP SIGNAL filed duration OFDM symbol interval τ = tSIFS Td = tSlot Δt = tSlot

Modulation BPSK BPSK QPSK QPSK 16-QAM 16-QAM 64-QAM 64-QAM

Code Rate 1/2 3/4 1/2 3/4 1/2 3/4 2/3 3/4

Data Rate 6 Mbps 9 Mbps 12 Mbps 18 Mbps 24 Mbps 36 Mbps 48 Mbps 54 Mbps

Bytes per Symbol 3 4.5 6 9 12 18 24 27

throughput from the source to the destination. However, when the channel gain increases, the throughput gain from the helper becomes less and less, and no helper can further increase the throughput when the received SNR is larger than 35 dB. Note that when the SNR is larger than 35 dB, the proposed scheme has a throughput slightly less than that in the IEEE 802.11a single-hop case, due to the overhead incurred in our scheme (i.e., the BTh busy-tone detection time). Fig. 4 illustrates the required average SNR versus throughput for the proposed scheme and single-hop transmission. To obtain a throughput of 11Mbps, our scheme requires an SNR of 3 dB SNR while the single-hop transmission requires an SNR larger than 10 dB. This means that transmit power can be greatly reduced in our scheme. However, the benefit reduces as the throughput increases. V. C ONCLUSION In this paper, we have proposed a novel cooperative triple busy tone multiple access scheme to facilitate node cooperation in wireless communications. With a cross-layer design

35 30

Throughput gain %

25 20 15 10 5 0 −5 0

5

10

15 20 25 Average SNR (dB)

30

IEEE 802.11a single−hop Proposed scheme

30 25 20 15 10 5

TABLE II PHY MODES IN IEEE 802.11A Mode 1 2 3 4 5 6 7 8

35

Average SNR (dB)

Parameter tSlot tSIFS tDIFS tPLCP Preamble tPLCP SIG tSymbol τ Td Δt

40

35

0 10

15

Fig. 4.

20 Throughput (Mbps)

25

30

Average SNR versus throughput

principle, an algorithm is presented to determine the capability of a node in helping transmissions between other nodes. With the use of BTh busy tone, an optimal helper can be chosen without disturbing existing transmissions. To achieve the same throughput, the proposed scheme requires much less transmit power than that in single-hop transmissions. Taking into account spatial frequency reuse in wireless networks, it is expected that the newly proposed scheme can further increase the network throughput by reducing the transmit power, which effectively reduces the interference among simultaneous transmissions in the network. R EFERENCES [1] J. N. Laneman and G. W. Wornell, “Distributed space-time coded protocols for exploiting cooperative diversity in wireless networks,” IEEE Trans. Inform. Theory, vol. 49, pp. 2415-2525, Oct. 2003. [2] J. N. Laneman, D. N. C. Tse, and G. W. Wornell, “Cooperative diversity in wireless networks: efficient protocols and outage behavior,” IEEE Trans. Inform. Theory, vol. 50, pp. 3062-3080, Dec. 2004. [3] M. Veluppillai, L. Cai, J.W. Mark, and X. Shen, “Maximizing cooperative diversity energy gain for wireless networks,” IEEE Trans. Wireless Commun., vol. 6, no. 7, pp. 2530-2539, July 2007. [4] M. Veluppillai, L. Cai, J.W. Mark, and X. Shen, “Partner selection based on optimal power allocation in cooperative diversity systems,” IEEE Trans. Veh. Technol., vol. 57, no. 1, pp. 511-520, Jan. 2008. [5] J. M. Shea, T.F. Wong, and W. H. Wong, “Cooperative-diversity slotted ALOHA,” Wireless Networks, vol. 13, pp. 361-369, June 2007. [6] P. Liu, Z. Tao, S. Narayanan, T. Korakis, and S. S. Panwar, “CoopMAC: a cooperative MAC for wireless LANs,” IEEE J. Selected Areas in Commun., vol. 25, pp. 340 - 354, Feb. 2007. [7] H, Zhu, G. Cao, “rDCF: A relay-enabled medium access control protocol for wireless ad hoc networks,” IEEE Trans. Mobile Comput., vol. 5, pp. 1201-1214, Sept. 2006. [8] Z. J. Haas, J. Deng, “Dual busy tone multiple access (DBTMA)-a multiple access control scheme for ad hoc networks,” IEEE Trans. Commun., vol. 50, pp. 975-985, June 2002. [9] P. Wang, H. Jiang, and W. Zhuang, “A dual busy-tone MAC scheme supporting voice/data traffic in wireless ad hoc networks,” in Proc. IEEE Globecom’06, Nov. 2006. [10] IEEE 802.11b, Part 11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications: High-Speed physical Layer Extension in the 2.4GHz Band, supplement to IEEE Standard, Sept. 1999. [11] IEEE 802.11a, Part 11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications: High-Speed physical Layer Extension in the 5GHz Band, supplement to IEEE Standard, Sept. 1999. [12] ETSI, “Broadband Radio Access Networks (BRAN); HIPERLAN type 2 technical specification; Physical (PHY) layer,” Dec. 2001. [13] C. Lee and L. H. C. Lee, Convolutional Coding: Fundamentals and Applications. Artech House Publishers, 1997. [14] J. G. Proakis, Digital Communications. McGraw-Hill, 4th ed., 2000.

40

Fig. 3. Throughput gain of our scheme compared with the IEEE 802.11a single-hop transmission
