Self-Regulating Network Utilization in Mobile Ad Hoc ... - IEEE Xplore

1302

IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 55, NO. 4, JULY 2006

Self-Regulating Network Utilization in Mobile Ad Hoc Wireless Networks Song Ci, Senior Member, IEEE, Mohsen Guizani, Senior Member, IEEE, Hsiao-Hwa Chen, Senior Member, IEEE, and Hamid Sharif, Member, IEEE

Abstract—In mobile ad hoc wireless LANs, it is very difficult to maintain a targeted network utilization due to the time-varying nature of the contention-based medium access control protocol and the lack of a central control. Furthermore, previous research has been mainly focusing on the aspect of optimizing the performance at each station. But doing so may result in a very low overall network utilization. Therefore, self-regulating network utilization is very important to provide quality-of-service (QoS) in mobile ad hoc wireless networks. Through self-disciplining its own behaviors locally, each station will optimize its protocol parameters to meet the targeted overall network utilization, which is very important for QoS provisioning to multimedia services. This paper proposes and evaluates a fully distributed scheme for each station to self-regulate its behaviors through adapting the local protocol parameters to meet the targeted overall network utilization with the changes in the network environment such as the number of stations and channel quality. Index Terms—Mobile wireless ad hoc network, network utilization, self-regulation.

I. I NTRODUCTION

R

ECENTLY, mobile ad hoc wireless LANs have been receiving more and more research efforts due to its rich application implications such as mobile wireless gaming, home entertainment, etc. It has been demonstrated by other researchers that there are many challenges making it very difficult to provide quality-of-service (QoS) under a mobile ad hoc wireless environment. In this paper, we will focus on one of the major challenges: how to make a wireless station selfregulate its own behaviors via adapting its medium access control (MAC) parameters to achieve a targeted overall network utilization. In this paper, network utilization is defined as the normalized goodput performance, standing for the average percentage of channel bandwidth used to transmit information data. Manuscript received October 15, 2005; revised November 9, 2005. The review of this paper was coordinated by Prof. X. Shen. S. Ci is with the Computer Science Department, University of Massachusetts, Boston, MA 02125 USA, and also with the Advanced Telecommunications Engineering Laboratory, University of Nebraska-Lincoln, Peter Kiewit Institute, Omaha, NE 68106 USA (e-mail: [email protected]). M. Guizani is with the Computer Science Department, Western Michigan University, Kalamazoo, MI 49008-5201 USA (e-mail: mguizani@cs. wmich.edu). H.-H. Chen is with the Institute of Communications Engineering, National Sun Yat-Sen University, Kaohsiung 804, Taiwan, R.O.C. (e-mail: hshwchen@ ieee.org). H. Sharif is with the Computer and Electronics Engineering Department, University of Nebraska-Lincoln, Lincoln, NE 68588 USA (e-mail: hsharif@ unl.edu). Digital Object Identifier 10.1109/TVT.2006.876687

Self-regulation toward a targeted overall network utilization at each wireless station is very important to service providers for carrying out a new multimedia service, which normally requires a desired overall network utilization. This means that more stations can be admitted into the network with a committed level of QoS if each of them can self-discipline its own behaviors. However, this is not an easy task. Although the widely deployed IEEE 802.11 standard becomes the de facto network standard of wireless LANs, the time-varying nature of ad hoc topology and carrier sense multiple access with collision avoidance (CSMA/CA) MAC protocol adopted by ad hoc wireless LANs makes it very difficult for each wireless station to self-regulate toward a target overall network utilization. Furthermore, in a mobile wireless environment, both the number of stations and the channel quality are dynamically changing, which makes the task even more challenging. Studies have shown that the performance of the CSMA/CA protocol is very sensitive to the number of stations and the channel quality in terms of frame error rate. In this paper, we will focus on how to make each wireless station self-regulate its behaviors to meet the targeted overall network utilization. Our solution to this problem is to develop a fully distributed scheme running at each wireless station in a mobile ad hoc wireless environment. Each station will dynamically change its MAC parameters such as data rate and contention window size with the changes of the network environment. Simulation results show that our proposed scheme will allow each station to self-regulate its MAC parameters for a predefined network utilization. The rest of this paper is organized as follows. We propose our scheme in Section II and discuss the simulation results in Section III. Section IV gives a brief review on related work. We conclude this paper in Section V.

II. S ELF -R EGULATING N ETWORK U TILIZATION A. System Model Before we start to formulate the problem of self-regulation for a targeted overall network utilization at each wireless station, let us first briefly review the system model used in our further study. In this paper, we assume that each station is a wireless terminal equipped with an IEEE 802.11 wireless LAN network interface. All wireless stations are organized as a mobile ad hoc wireless network, i.e., there are no central control devices such as access points existing in the network. Ad hoc

0018-9545/$20.00 © 2006 IEEE

CI et al.: SELF-REGULATING NETWORK UTILIZATION IN MOBILE Ad Hoc WIRELESS NETWORKS

1303

and Tc must be measured in the number of time slots. Note that (1) is valid for both basic and request-to-send/clear-tosend (RTS/CTS) access methods. When using it to calculate the throughput performance of the basic access method, we can simply take out all parameters related to RTS/CTS. In (1), Lxxx is the length of different types of frames, where xxx could be one of the following frame types: RTS, CTS, data, and ACK. Pb is the bit error rate. Pexxx is the frame error rate of frame type xxx. Furthermore, we can calculate the overhead caused by frame errors, where Texxx is the overhead caused by the frame errors of frame type xxx. In a noisy channel, the probabilities of frame errors have been derived in [1]. Let H = PHYhdr + MAChdr be the packet header and δ be the propagation delay. Then we have the following equations for Ts and Tc . For the RTS/CTS access method, c ollisions can occur only on RTS frames, and thus, we obtain Tsrts/cts = DIFS + RTS + CTS + H + E[P ] + 4δ + 3 ∗ SIFS + ACK Tcrts/cts Fig. 1.

Wireless ad hoc networks.

routing algorithms could be used to allow any pair of stations to talk to each other, but it is out of the scope of this paper. Fig. 1 shows a typical network topology of wireless ad hoc networks adopted in this paper, where six wireless stations are organized as a peer-to-peer ad hoc wireless network. Because each station uses a wireless LAN network interface running a CSMA/CA MAC protocol, we also adopt the Markov chain model to analyze the performance of network utilization, as shown in Fig. 2, where retry limit, back-off suspensions, and fading channel errors are taken into considerations. Consider an ad hoc network with n contending stations and define contention window size Wi at back-off stage i. For convenience, Wi = 2i W0 , where i ∈ [0, m] is called the back-off stage i, W0 is the initial contention window size, and m is the maximum back-off stage. As shown in [1], network utilization Λ under a fading channel can be expressed by (1), shown at the bottom of the page, where E[P ] is the average packet length, Ts is the average time that the channel is sensed busy because of a successful transmission, and Tc is the average time that the channel is sensed busy by the stations during a collision. σ is the duration of an empty slot time. The times E[P ], Ts

Λ=

= DIFS + RTS + δ

(2)

where E[P ∗ ] is the average length of the largest packet payload involved in a collision. If we have a case where all packets have the same fixed size, then E[P ∗ ] = E[P ] = P . The probability τ that a station transmits in a randomly chosen slot can be expressed as τ=

m

bi,0 =

i=0

1 − pm+1 b0,0 1−p

where the value b0,0 can be expressed as that in (4), shown at the bottom of the page, and the transmission error rate p is p = 1 − (1 − τ )n−1 (1 − Pe ).

τ =1−

1−p 1 − Pe

1 n−1

.

(

(6)

More details about the Markov model under a fading channel can be found in [1].

 2(1−2p)(1−p)   W (1−(2p)m+1 (1−p)+(1−2p)(1−pm+1 ) ,  W

(5)

Equations (3) and (5) represent a nonlinear system with two unknown variables τ and p, which can be solved by using numerical techniques. So, we will obtain τ as

Ptr Ps (1 − Pe )E[P ∗ ] (1 − Ptr )σ + Ptr Ps Ts (1 − Pe ) + Ptr (1 − Ps )Tc + Ptr Ps (Perts Terts + Pects Tects + Pedata Tedata + Peack Teack )

b0,0 =

(3)

2(1−2p)(1−p)

1−(2p)m +1 (1−p)+(1−2p)(1−pm+1)+W 2m pm +1 (1−2p)(1−pm −m )

(1)

m ≤ m , m > m

(4)

1304


Fig. 2. Markov chain model of the IEEE 802.11 DCF.

In this paper, we also assume that each station will have means to learn the changes in a network environment such as the number of stations and the channel quality. It is true that most of the PHY and MAC parameters will be available in most of the commercial implementations of IEEE 802.11 wireless LANs.

J = Φ [x(N )] +

N −1

L [x(i), u(i), i] .

(8)

0

In most real systems, there is a set of constraint relations c(i), which is defined as

B. Problem Formulation As discussed above, a mobile ad hoc wireless LAN is highly dynamic in terms of network topology, channel quality, and number of stations. This allows us to characterize an ad hoc wireless LAN as a discrete dynamic system, which can be described by an n-dimensional state vector x(i) at step i, and the state x(i + 1) is determined by an m-dimensional control vector u(i), expressed as [2] x(i + 1) = f [x(i), u(i), i]

u(i) for i = 0, 1, . . . , N − 1 to minimize a performance index of the form

(7)

x(0) = x0 , which is the initial state of a dynamic system. Then, we are trying to find the sequence of control vectors

c(i) (u) = 0,

i = 1, . . . , n.

(9)

If there are inequality constrains, we normally express them as c = ATu − b < 0.

(10)

According to the method of Lagrange multipliers, we can derive H = L + λTc .

(11)

Here, H is defined as the discrete Hamiltonian H(i) ≡ H[x(i), u(i), λ(i + 1), i], where all variables are n-dimensional vectors and λi is called a Lagrange multiplier.


So far, we have derived all the relations needed to solve the unknown control vector u(i). Given an initial guess of u(i), the state vector x(i) can be calculated to determine J. Then, we can iterate this procedure to determine the control histories u(i) for i > 0. Thus, Ju(i) , which is the gradient of J with respect to u(i), can be calculated numerically by the state histories x(i). Therefore, we have J = Φ [x(N )] +

N −1

1305

TABLE I WIRELESS LAN PARAMETERS USED IN SIMULATIONS

L(i) + λT (i + 1)

0

× [f (i) − x(i + 1)] λT (0) [x0 − x(0)] . (12) Then we consider differential changes in J with respect to u(i) and x0 and derive λT (i) = Hx (i) ≡ Lx (i) + λT (i + 1)fx (i).

(13)

After discussing the generic dynamic system, let us define the self-regulated network utilization Φ[x(N )] = Λ(N ), which is a function of the state vector x(i) and the control vector u(i). In this paper, L = 0, x(i) is the network utilization at time i, and u(i) is the control vector composed with data rate (α) and the transmission probability that each station transmits in a time slot (τ ), i.e., u = [α, τ ] Λ(i) = L (x(i), u(i), i) .

(14) III. N UMERICAL S TUDY

According to the network utilization equation described in Section II, which is a nonlinear function. Furthermore, the following constraints have to be followed: c=

αmin ≤ α ≤ αmax τ min ≤ τ ≤ τ max

.

(15)

So far, we have formulated our proposed scheme as a nonlinear optimization problem with a set of linear constraints. In most practical systems, the target network utilization θ is known for a certain type of application. Therefore, the proposed scheme is to make the network utilization be as close as possible to its committed (overall) network utilization. Recall that our goal in this paper is that each station self-regulate itself by using the appropriate protocol parameters to achieve the targeted overall network utilization. Thus, we can define the quadratic performance index to meet our goal as J=

1 (Λ(i) − θ)2 . 2

protocol to adapt its key parameters to meet the targeted overall network utilization according to the changes of network load and channel quality.

(16)

Then we will minimize the quadratic performance index J with respect to the control vector u given the state vector x. Before each transmission or retransmission, the proposed self-regulation scheme will run at each station to derive a set of the control vector u, which will be used for the next transmission. This new control vector will allow the CSMA/CA

Simulations have been set up for evaluating the performance of the proposed self-regulation scheme. Our goal here is to evaluate the performance of the proposed scheme compared with that achieved by the conventional fixed scheme under various network scenarios in terms of channel quality and network load. The IEEE 802.11 MAC is simulated by implementing the Markov model discussed in Section II. Table I lists all the parameters of the PHY and MAC layers used in our simulations. As previously mentioned, a mobile ad hoc wireless environment has been considered in this paper. Because most movements in a mobile ad hoc wireless LAN are at a slow speed, a pedestrian movement is set to be 1 m/s in this paper. We also only consider the slow-fading channel, where the Doppler frequency fd is 10 Hz. Therefore, Jakes’ model is used to generate the fading pattern, resulting in a fluctuating receiving signal-tonoise ratio (SNR) for our simulations. Fig. 3 shows the fading channel used in our experimental study, where fd is 10 Hz. In this paper, a worse channel quality with 3 dB less is also considered to evaluate our proposed scheme comprehensively. For the proposed self-regulation scheme, the range of data rate is from 1 to 11 Mb/s, and the range of τ is from 0.008 to 0.1, approximately equivalent to CWmin = 255 and CWmin = 7. The initial control vector is set to 11 Mb/s and 0.008, corresponding to the most efficient combination of MAC parameters. Table II shows the performance of the proposed selfregulation scheme with different network sizes, where the targeted overall network utilization θ is 0.7. From Table II,

1306


Fig. 3. Fading channel (fd = 10 Hz).

Fig. 5. Optimal τ . The network size is 20 stations, and the targeted overall network utilization is 0.7.

TABLE II NETWORK UTILIZATIONS ACHIEVED BY THE PROPOSED SELF-REGULATING SCHEME WITH θ = 0.7

Fig. 6. Optimal CWmin. The network size is 20 stations, and the targeted overall network utilization is 0.7.

Fig. 4. Optimal data rate. The network size is 20 stations and the targeted overall network utilization is 0.7.

we can observe that the proposed self-regulating scheme works well in all cases, especially when the network size is larger, which is actually a desired feature because self-regulation is really needed when a network has more stations. Notice that the average overall network utilization decreases when the number of stations increases, although the median is the targeted overall network utilization. This tells us the proposed self-regulating algorithm still depends on the network size. Thus, some mechanisms such as admission control have to be used with the

proposed scheme to deliver a desired QoS. Figs. 4–6 reflect the performance of the proposed self-regulating scheme, where the network size is 20 stations and the targeted overall network utilization is 0.7. Fig. 4 shows the data rate adaptation made by a station to meet the targeted overall network utilization, where the initial data rate is gradually downgraded to 1 Mb/s due to the relatively large number of stations. Fig. 5 shows the self-regulation in terms of transmission probability in a slot τ . Depending on the channel quality, each station will choose the optimal transmission probability τ to achieve the targeted overall network utilization through adapting its CWmin, which is shown in Fig. 6. A. Further Discussions In the following sections, we will study the performance of the proposed scheme from different perspectives to gain an in depth understanding of the proposed scheme.


Fig. 7. Optimal data rate. The network size is five stations, and the targeted overall network utilization is 0.7.

Fig. 8. Optimal τ . The network size is five stations, and the targeted overall network utilization is 0.7.

1) Number of Stations: We simulated the performance of the proposed self-regulating scheme with different network sizes in terms of the number of stations in the network. Figs. 7–9 show the self-regulation with five stations in a network, corresponding to data rate, τ , and CWmin, respectively. Similarly, Figs. 10–12 are derived with 50 stations in a network. From the simulation results, we can observe that the average data rate is higher, the transmission probability is lower, and CWmin is smaller when the network size is small, compared with a large network. We should also notice that although the average overall network utilization is higher when the network size is small, its variation is larger. This is because in a small network, every network activity will produce a large impact on other stations, which is generally not the case in a large network. In other words, the decisions made by each station will not affect others that much in a large network. 2) Channel Quality: We also simulated the impacts of channel quality on the performance of the proposed self-regulating scheme. The channel quality used in the following simulations

1307

Fig. 9. Optimal CWmin. The network size is five stations, and the targeted overall network utilization is 0.7.

Fig. 10. Optimal data rate. The network size is 50 stations, and the targeted overall network utilization is 0.7.


1308

Fig. 12. Optimal CWmin. The network size is 50 stations, and the targeted overall network utilization is 0.7.


Fig. 14. Optimal τ under a bad channel quality. The network size is 20 stations, and the targeted overall network utilization is 0.7. TABLE III NETWORK UTILIZATIONS ACHIEVED BY THE PROPOSED SELF-REGULATING SCHEME WITH θ = 0.5

Fig. 13. Optimal data rate under a bad channel quality. The network size is 20 stations, and the targeted overall network utilization is 0.7.

are 3 dB worse than that shown in Fig. 3. Fig. 13 shows the data rate adaptation using the proposed scheme, and Fig. 14 shows the adaptation of τ using the proposed scheme. From the simulation results, each station will reduce its transmission probability by adopting more conservative MAC parameters such as data rate and CWmin to meet the targeted overall network utilization. 3) Targeted Network Utilization: We also simulated the performance of the proposed scheme with different targeted overall network utilization. Table III and Figs. 15 and 16 show the performance of the proposed scheme with a lower targeted overall network utilization θ = 0.5. From these simulation results, we can observe that the proposed scheme performs well with a lower targeted network utilization. Notice that in Table III both the average and the median of overall network utility deviated largely from the targeted network utility. This anomaly indicates that there exists an adaption limit, although the proposed scheme performs well in a broad range of network size and channel quality. In this case, the actual overall network


utilization is much higher than the targeted one, which is over the capacity of the proposed scheme. Therefore, we can conclude that the proposed self-regulation scheme allows a wireless station to self-regulate its MAC parameters to meet the targeted overall network utilization as close as possible. IV. R ELATED W ORK For modeling the back-off process of the 802.11 DCF, many research efforts have produced models of the IEEE 802.11


Fig. 16. Optimal τ . The network size is 50 stations and the targeted overall network utilization is 0.5.

MAC protocol. In [3] and [4], the performance of the asynchronous data transfer protocols has been evaluated through extensive numerical and simulation results, following the specifications of the IEEE 802.11 standard. Some other issues such as the possibility of capture and the presence of hidden stations have been discussed in that paper as well. In [5], the p-Persistent model has been proposed to characterize the MAC layer of IEEE 802.11 wireless LANs. Recently, Markov chain has been introduced to accurately model the back-off procedure in [1], [6], and [7]. Link adaptation has been proposed for improving system performance by adaptively changing the protocol parameters according to channel quality and network load. In [8]–[11], the optimal frame size prediction has been studied. The basic idea is to get the maximum throughput by dynamically changing the frame size according to the variations of channel quality. When the channel quality is good, a larger frame size can be used to get a higher throughput performance. A shorter frame size is otherwise adopted to lower the number of retransmissions. In [12], link adaptation has been addressed from several aspects such as frame size, equalizer design, and power control in a Rayleigh-fading environment. Recently, algorithms of fragmentation threshold adaptation have been proposed and discussed. When interferences appear, a long frame could be divided into several short fragments at the transmitters, and they will be reassembled at the receivers after all fragments are correctly received. In [13], optimal contention window size algorithms were proposed and discussed. It showed that the throughput performance could be changed significantly by optimizing protocol parameters. Adaptive modulation schemes over fading channels have been studied extensively in the literature [14], [15].

V. C ONCLUSION QoS provisioning in a mobile ad hoc wireless environment usually requires the overall network utilization to be maintained

1309

at a certain level. However, the dynamic natural of the mobile ad hoc wireless environment and the lack of central control make this very difficult. Previous research has been mainly focusing on the aspect of optimizing the performance at each station. But doing so may not have a positive impact on the overall network utilization. In other words, if each station only does optimizations on its own behalf, the overall network utilization could be very low. In this paper, we have addressed an important issue of how to make each station in a mobile ad hoc wireless environment optimize its MAC parameters not only on its own behalf but also for the overall network utilization. We have proposed a full distributed scheme to allow each station to self-regulate its behaviors through adapting its protocol parameters for the targeted network utilization. We have formulated this problem to be a dynamic optimization problem with linear constraints. Theoretical analysis has been conducted and extensive simulations have been conducted to evaluate the proposed algorithm. The simulation results show that the proposed scheme will allow each station to self-regulate to meet the targeted network utilization by adapting its key system parameters of PHY and MAC layers according to the changes of network load and channel quality. R EFERENCES [1] S. Ci, H. Sharif, and P. Mahasukhon, “Evaluating saturation throughput performance of the IEEE 802.11 MAC under fading channels,” in Proc. IEEE/ACM BroadNets Wireless Netw. Symp., 2005, pp. 726–731. [2] A. E. Bryson, Dynamic Optimization. Englewood Cliffs, NJ: PrenticeHall, 1999. [3] H. Chhaya and S. Gupta, “Performance modeling of asynchronous data transfer methods of IEEE 802.11 MAC protocol,” Wireless Netw., vol. 3, no. 3, pp. 217–234, 1997. [4] F. Eshghi and A. Elhakeem, “Performance analysis of ad hoc wireless LANs for real-time traffic,” IEEE J. Sel. Areas Commun., vol. 21, no. 2, pp. 204–215, Feb. 2003. [5] F. Cali, M. Conti, and E. Gregori, “Dynamic tuning of the IEEE 802.11 protocol to achieve a theoretical throughput limit,” IEEE/ACM Trans. Netw., vol. 8, no. 6, pp. 785–799, Dec. 2000. [6] G. Bianchi, “Performance analysis of the IEEE 802.11 distributed coordination function,” IEEE J. Sel. Areas Commun., vol. 18, no. 3, pp. 535–547, Mar. 2000. [7] H. Wu et al., “Performance of reliable transport protocol over IEEE 802.11 wireless LAN: Analysis and enhancement,” in Proc. IEEE INFOCOM, 2002, pp. 599–607. [8] S. Ci, H. Sharif, and A. Young, “Frame size prediction for indoor wireless network,” Electron Lett., vol. 37, no. 18, pp. 1135–1136, 2001. [9] E. Modiano, “An adaptive algorithm for optimizing the packet size used in wireless ARQ protocols,” Wireless Netw., vol. 5, no. 4, pp. 279–286, Jul. 1999. [10] P. Lettieri and M. B. Srivastava, “Adaptive frame length control for improving wireless link throughput, range, and energy efficiency,” in Proc. IEEE INFOCOM, 1998, vol. 2, pp. 564–571. [11] D. Qiao, S. Choi, and K. Shin, “Goodput analysis and link adaptation for IEEE 802.11a wireless LANs,” IEEE Trans. Mobile Comput., vol. 1, no. 4, pp. 278–292, Oct.–Dec. 2002. [12] C. Chien, M. Srivastava, R. Jain, P. Lettieri, V. Aggarwal, and R. Sternowski, “Adaptive radio for multimedia wireless links,” IEEE J. Sel. Areas Commun., vol. 17, no. 5, pp. 793–813, May 1999. [13] F. Cali, M. Conti, and E. Gregori, “Dynamic tuning of the IEEE 802.11 protocol to achieve a theoretical throughput limit,” IEEE/ACM Trans. Netw., vol. 8, no. 6, pp. 785–799, Dec. 2000. [14] X. Qiu and K. Chawla, “On the performance of adaptive modulation in cellular systems,” IEEE Trans. Commun., vol. 47, no. 6, pp. 884–895, Jun. 1999. [15] D. Goeckel, “Adaptive coding for time-varying channels using outdated fading estimates,” IEEE Trans. Commun., vol. 47, no. 6, pp. 844–855, Jun. 1999.

1310

Song Ci (S’99–M’02–SM’06) received the B.S. degree from the Shandong University of Technology (now Shandong University), Jinan, China, in 1992, the M.S. degree from the Chinese Academy of Sciences, Beijing, China, in 1998, and the Ph.D. degree from the University of Nebraska-Lincoln in 2002, all in electrical engineering. He was a Telecommunication Engineer with China Telecom, Shandong, China, from 1992 to 1995 and an R&D Co-Op with the Wireless Connectivity Division, 3COM Cooperation, Santa Clara, CA, in 2001. From 2002 to 2005, he was an Assistant Professor of computer science at the University of Michigan-Flint. He is an Assistant Professor of computer science at the University of Massachusetts, Boston. He is also with the Advanced Telecommunications Engineering Laboratory, University of Nebraska-Lincoln, Peter Kiewit Institute, Omaha, NE. He has published more than 30 technical papers in referred journals and international conferences in these areas. His current research interests include cross-layer design for QoS provisioning in wireless data networks, wireless sensor networks, and 3G/4G cellular networks; heterogeneous wireless data network interconnection; and low-power embedded real-time system design. Dr. Ci is a member of the Association for Computing Machinery. He currently serves as the Associate Editor in the Editorial Board of Wireless Communications and Mobile Computing. He is also a Reviewer for many refereed journals and a Technical Committee Member for several international conferences. He received the Best Paper Award in the 2004 IEEE International Conference on Networking, Sensing, and Control.

Mohsen Guizani (S’87–M’90–SM’98) received the B.S. (with distinction) and M.S. degrees in electrical engineering and the M.S. and Ph.D. degrees in computer engineering from Syracuse University, Syracuse, NY, in 1984, 1986, 1987, and 1990, respectively. He is currently a Professor and the Chair in the Computer Science Department, Western Michigan University, Kalamazoo. He is the author of three books and in the process of writing another two. He has more than 120 publications in refereed journals and conferences. His research interests include computer networks, wireless communications and computing, and optical networking. Dr. Guizani is a member of the IEEE Communication Society, the IEEE Computer Society, American Society for Engineering Education, Association for Computing Machinery, Optical Society of America, SCS, and Tau Beta Pi. He currently serves on the editorial boards of six technical journals and is the Founder and EIC of Wireless Communications and Mobile Computing (http://www.interscience.wiley.com/jpages/ 1530-8669/). He guest edited in a number of special issues in journal and magazines. He also served as Member, Chair, and General Chair of a number of conferences. He received both the Best Teaching Award and the Excellence in Research Award from the University of Missouri-Columbia in 1999 (a college-wide competition). He won the Best Research Award from the King Fahd University of Petroleum and Minerals (KFUPM) in 1995 (a university wide competition). He was selected as the Best Teaching Assistant for two consecutive years at Syracuse University in 1988 and 1989.


Hsiao-Hwa Chen (SM’01) received the B.Sc. and M.Sc. degrees from Zhejiang University, Zhejiang, China, in 1982 and 1985, respectively, and the Ph.D. degree from the University of Oulu, Oulu, Finland, in 1990, all in electrical engineering. He was with the Academy of Finland for research on spread-spectrum communications as a Research Associate during 1991–1993 and the National University of Singapore as a Lecturer and then a Senior Lecturer from 1992 to 1997. He joined the Department of Electrical Engineering, National Chung Hsing University, Taichung, Taiwan, R.O.C., as an Associate Professor in 1997 and was promoted to a Full Professor in 2000. He was a Visiting Professor with the Department of Electrical Engineering, University of Kaiserslautern, in 1999, the Institute of Applied Physics, Tsukuba University, in 2000, and the Institute of Experimental Mathematics, University of Essen, in 2002. In 2001, he joined the National Sun Yat-Sen University, Kaohsiung, Taiwan, as the Founding Director of the Institute of Communications Engineering. Under his leadership, the institute was ranked second place in the country in terms of SCI journal publications and National Science Council funding per faculty in 2004. He has authored or coauthored over 120 technical papers in major international journals and conferences, and three books and two book chapters in the areas of communications. Dr. Chen served as TPC Member and Symposium Chair of major international conferences, including IEEE VTC, IEEE ICC, IEEE GLOBECOM, etc. He served or is serving as a member of the Editorial Board and Guest Editor of IEEE Communications Magazine, IEEE JOURNAL ON SELECTED AREAS IN C OMMUNICATIONS , Wireless Communications and Mobile Computing Journal, International Journal of Communication Systems, etc. He has been a Guest Professor with Zhejiang University since 2003. He was the recipient of numerous Research and Teaching Awards from the National Science Council and Ministry of Education, Taiwan, from 1998 to 2001.

Hamid Sharif (S’82–M’97) received the B.S., M.S., and the Ph.D. degrees from the University of Iowa, Iowa City, the University of Missouri-Columbia, and the University of Nebraska-Lincoln, all in electrical engineering. He is a Professor in the Computer and Electronics Engineering Department, Peter Kiewit Institute, University of Nebraska-Lincoln. He is the current director of the Advanced Telecommunications Engineering Laboratory in the Peter Kiewit Institute. His research work has been published in many journal papers and numerous conference papers. His research areas include wireless communications networks, wireless sensor networks, and QoS in IP networks.