A Distributed OFDMA Medium Access Control for Underwater Acoustic Sensors Networks

Transcription

1 This full text paper was peer reviewed at the diretion of IEEE Communiations Soiety subjet matter experts for publiation in the IEEE ICC 2011 proeedings A Distributed OFDMA Medium Aess Control for Underwater Aousti Sensors Networks Fatma Bouabdallah and Raouf Boutaba Shool of Computer Siene, University of Waterloo ; 200 University Ave. W., Waterloo, ON, Canada {fbouabda,rboutaba}@uwaterloo.a Abstrat In this paper, we propose UW-OFDMAC, a distributed Medium Aess Control (MAC protool tuned for UnderWater Aousti Sensor Networks (UW-ASNs. It is a transmitter-based Orthogonal Frequeny Division Multiple Aess (OFDMA sheme that integrates an original power and OFDMA parameters self-assignment algorithm to set the optimal transmit power, subarrier spaing and guard interval duration. UW- OFDMAC aims at ahieving two objetives, namely, guarantee high bandwidth effiieny and low energy onsumption. Simulation results show that UW-OFDMAC outperforms the basi OFDMA protool tuned for the underwater environment. Index Terms Underwater Aousti Sensor Networks, energy onservation, Bandwidth Effiieny, OFDMA protool, ollision free aess. I. INTRODUCTION Two thirds of the earth surfae onsist of water. Most of whih remain unexplored. Inreasingly researh efforts are devoted to Underwater Aousti Sensor Networks due to their important appliations for military and ommerial purposes. Broad appliations of UNDERWATER sensor networks inlude among others: offshore exploration, tsunami warning, and mine reonnaissane [1]. Aousti ommuniation is the most suitable physial layer tehnology for underwater networks. Indeed, eletromagneti waves propagate through ondutive salty water only at very short range due to the high attenuation and absorption effet in underwater environment. Optial waves do not suffer from suh high attenuation but are affeted by sattering. Thus links in underwater networks are usually based on aousti wireless ommuniations, whih pose unique hallenges due to the harsh underwater environment suh as limited bandwidth [2], high and variable propagation delays [3], high bit error rates and temporary losses of onnetivity aused by multipath and fading phenomena [4], and asymmetri links. Coneiving a Medium Aess Control (MAC protool is a major hallenge for the deployment of UnderWater Aousti Sensor Networks (UW-ASNs [1]. Ideally, an optimal underwater MAC protool should provide high network throughput, and low energy onsumption, taking into aount the harsh harateristis of the underwater propagation medium. We believe that Orthogonal Frequeny Division Multiple Aess (OFDMA is the most promising physial layer and multiple aess tehnique for UW-ASNs. Reall that, OFDMA is a speial ase of multiarrier modulation in whih multiple user symbols are transmitted simultaneously using different subarriers with overlapping frequeny bands that are mutually orthogonal. The inner nature of the OFDMA protool makes it the most appropriate tehnology for underwater ommuniations sine i it is robust to frequeny-seletive fading due to subarriers orthogonality, ii ompensates for the effet of multipath by inserting guard interval between the OFDMA symbols, thus eliminating the Inter-Symbol Interferene, and iii allows simultaneous ollision free transmissions between neighboring nodes. As a result, OFDMA enables the most effiient use of the sare aousti spetrum and redues paket retransmissions, whih results in dereased energy onsumption and inreased network throughput. For these reasons, in this paper we propose UW-OFDMAC, a transmitter-based OFDMA MAC protool for UW-ASNs that integrates an original power and OFDMA parameters selfassignment algorithm so as to maximize bandwidth effiieny while minimizing energy onsumption. The objetive of UW- OFDMAC is twofold: i guarantee high bandwidth effiieny and ii guarantee low energy onsumption. We prove that UW-OFDMAC manages to simultaneously ahieve the two objetives in shallow water ommuniations, whih may be heavily affeted by multipath fading. We also formulate and resolve the distributed power and OFDMA parameters self-assignment problem to maximize bandwidth effiieny while providing ollision free aess. UW-OFDMAC uses loally generated OFDMA parameters to spread transmitted signals on the available bandwidth, whih guarantees high bandwidth effiieny. To the best of our knowledge, UW-OFDMAC is the first protool that leverages OFDMA properties to ahieve multiple aess in the bandwidth-limited underwater hannel. Previous works [5] [6] onsidered OFDMA shemes from a physial layer perspetive. The main features that haraterize UW- OFDMAC are: i it is fully distributed, sine OFDMA optimal parameters and transmit power are distributively seleted by eah transmitter without relying on a entralized entity; and ii it guarantees low energy onsumption and high bandwidth effiieny while avoiding interferenes. The remainder of this paper is organized as follows. In Setion II, we disuss the suitability of existing ad ho and sensor MAC protools for the underwater environment. Setion III presents the OFDMA system studied in this paper. In setion IV, we introdue UW-OFDMAC, and in Setion V we formulate and resolve the distributed power and OFDMA parameters self-assignment problem. In Setion VI, we ompare through simulation UW-OFDMAC with a basi OFDMA sheme tuned for the underwater environment. Finally, Setion VII onludes this paper. II. RELATED WORK There has been intensive researh on MAC protools for ad ho [7] and wireless terrestrial sensor networks [8] in the last deade. However, due to the different nature of the underwater environment and appliations, existing terrestrial MAC solutions are unsuitable for this environment. In fat, hannel aess ontrol in UW-ASNs poses additional hallenges due to the peuliarities of the underwater hannel, in partiular limited bandwidth, very high and variable propagation delays, high bit error rates, temporary losses of onnetivity, hannel /11/$ IEEE

2 This full text paper was peer reviewed at the diretion of IEEE Communiations Soiety subjet matter experts for publiation in the IEEE ICC 2011 proeedings asymmetry, and heavy multipath and fading phenomena. For a thorough disussion on the reasons why several multiple aess tehniques widely employed in terrestrial sensor networks suh as TDMA, FDMA, and CSMA, are not suitable for the underwater environment, we refer the reader to [1]. Here, we mainly onentrate on previous work on CDMA and OFDMA, sine they are attrating more and more the attention of the UW-ASNs researh ommunity. A distributed CDMA-based energy-effiient MAC protool for the underwater environment was reently proposed in [9]. It is a transmitter-based Code Division Multiple Aess (CDMA sheme alled UW-MAC. It integrates a losed-loop distributed algorithm to optimize the transmit power and ode length to mitigate the near-far problem. UW-MAC aims at guaranteeing high network throughput with low hannel aess delay, and low energy onsumption. However, UW-MAC requires that all nodes have knowledge of all other nodes multiple aess interferene (MAI whih may be restritive for energy onstrained sensor devies. Moreover, in CDMA based systems, multipath fading is ombated leveraging RAKE filters at the reeivers. Rake reeivers typially detet a fixed number of multipath omponents and oherently ombine them. Besides, the CDMA interferene an be mitigated by means of equalizer whih inreases the system omplexity. Orthogonal Frequeny Division Multiple Aess (OFDMA is one of the reently proposed methods. It is the primary hoie for future broadband ommuniation networks (WiMAX and IEEE n. The main feature of the OFDMA protool is its robustness to multipath fading. In OFDMA system, the number of multi-path omponents does not limit the performane of the system as long as all these multipaths are within the guard interval duration. Therefore, in underwater wireless systems where multipath effet is very ommon, OFDMA systems are more robust and less omplex than CDMA systems. A few researh ativities applied OFDMA to underwater aousti appliations [5] [6]. Most of these works fous on the modulation aspet and not on the multiple-aess aspet. In other words, OFDMA has been mostly regarded as a modulation tehnology. Hayajneh. et. al reently proposed an OFDMA based MAC protool for underwater aousti sensor networks [10]. The main idea of this work is to arefully assign the available subarriers among neighboring nodes so that ollision is ompletely avoided. However, in our work, we assume that eah sensor devie is assigned a subhannel and the OFDMA parameters as well as the transmission power are determined eah time a new transmission is triggered. In other words, eah sensor an dynamially adapt OFDMA parameters depending on the reeiver loation and motion. The main objetive of our work is to guarantee high bandwidth effiieny and low energy onsumption while providing ollision free aess. III. OFDMA PARAMETERS Orthogonal Frequeny Division Multiplexing (OFDM is a multiplexing tehnique that subdivides the available bandwidth into multiple orthogonal frequeny sub-arriers. The input data stream is divided into several parallel sub-streams of redued data rate (inreased symbol duration and eah sub-stream is transmitted on a separate orthogonal sub-arrier. In an OFDM system, resoures are available in the time domain by means of OFDM symbols and in the frequeny domain by means of subarriers. The time and frequeny resoures an be organized into sub-hannels for alloation to individual users. OFDMA is a multi-user version of OFDM digital modulation sheme. Multiple aess is ahieved in OFDMA by assigning subsets of subarriers to individual users. This allows simultaneous low data rate transmission from several users. A. Sub-Carrier Spaing : Δf In OFDMA, the sub-arrier frequenies should be hosen so that the sub-arriers are orthogonal to eah other, meaning that ross-talk between the sub-arriers is eliminated or in other words Inter-Carrier Interferene (ICI is avoided. The orthogonality requires that the sub-arrier spaing Δf is muh larger than the Doppler Shift whih is mainly aused by the sender and reeiver motion. In other words, the Doppler shift is the soure of frequeny deviation between the transmitter and the reeiver. An OFDMA system an only tolerate a frequeny offset that is muh smaller than the arrier spaing. Any residual frequeny offset auses loss of orthogonality between the arriers, and the resulting interarrier interferene (ICI leads to performane degradation. B. Guard Interval Tg One key priniple of OFDMA is to transmit a number of low-rate streams in parallel instead of a single high-rate stream. By doing so, Inter-Symbol Interferene aused by multipath is redued. Consequently, sine the duration of eah symbol is long, it is feasible to insert a guard interval between the OFDMA symbols, thus eliminating the Inter-Symbol Interferene (ISI. By fixing the subarrier spaing and guard interval duration, the basi unit of physial resoure (time and frequeny is fixed. One immediate advantage stemming from OFDMA parametrization is the flexibility of deployment. In fat, OFDMA parameters an be adapted aording to the reeiver loation and motion in order to maximize the bandwidth effiieny. Indeed, the bandwidth effiieny an be highly improved with losely loated reeiver moving with the same veloity and through the same diretion as the sender. CDMA based systems however, do not provide suh a natural flexibility. IV. UW-OFDMAC: AN OFDMA MAC FOR UW-ASNS Our proposed UW-OFDMAC protool aims at setting the optimal ombination of transmit power, subarrier spaing and guard interval duration at the transmitter side so as to minimize the energy onsumption and ompletely mitigate the ICI and ISI effets. For this purpose, we assume that eah user is initially assigned a sub-hannel (onseutive set of sub-arriers. Note that, in order to avoid Multi-user Aess Interferene the whole set of sub-hannels must be mutually exlusive. To illustrate, assume that M users share the available bandwidth whih is evenly divided into M sub-hannels. I m represents the m th sub-hannel assigned to user m. Clearly to avoid that a given sub-hannel is shared by different users, the sets {I m } must be mutually exlusive, i.e., I m I j = for m j. The sub-hannel assigned to eah user an be subdivided into multiple orthogonal frequeny sub-arriers. Therefore, with N sub-arriers, the total passband bandwidth of user m will be I m N Δf(Hz. Note that the total number of sub-arriers for eah user (N mainly depends on the reeiver loation and motion. In other words, eah user an dynamially adapt the

3 This full text paper was peer reviewed at the diretion of IEEE Communiations Soiety subjet matter experts for publiation in the IEEE ICC 2011 proeedings OFDMA parameters (in terms of sub-arrier spaing and guard interval duration as well as the transmission power depending on the reeiver loation and motion. Consequently, we highly improve the resulting bandwidth and energy effiieny while minimizing users interferene. In fat, nearer reeivers need muh less power onsumption in order to suessfully deliver the data pakets. Moreover, with near reeivers, the guard interval an be highly redued while avoiding the inter-symbol interferene hene improving the total network throughput. Adopting suh sub-arrier alloation strategy imposes that the sub-arrier frequeny spaing hosen by a given transmitter should be ommuniated to the reeiver in order to adjust its window to the transmitter sub-arriers number. In UW-OFDMAC protool, a sensor node randomly aesses the hannel transmitting a notifiation paket (NP, whenever a new transmission is triggered. The NP of size L NP symbols, is sent using ommon OFDMA parameters known by all devies (Δf = 1 T,Tg = dmax, where T is the symbol duration, d max is the maximal inter-node distane and = 1500m/s is the sound veloity of the hannel. Sender i sends to its next hop j, d ij meters distant, the notifiation paket NP. The NP inorporates information about the final destination (i.e., the surfae station, the hosen next hop (i.e., node j, and the OFDMA parameters (,Tg ij that i will use to transmit the atual data paket of size L D =(p symbols. Taking advantage of the reeived information, node j will be able to synhronize with the signal from i by adjusting its window to user i sub-arriers number. Immediately after the transmission of the NP, i transmits the data paket on the hannel using the optimal transmit power Ptij, and Tg ij set by the power and OFDMA parameters self-assignment algorithm. One j has orretly reeived the data paket from i, an ACK paket is immediately sent to aknowledge the suessful reeption. Note that, ACK paket of size L A = (q Ij Δf symbols is sent using Ptij, and Tg ij. In ase i does not reeive the ACK before the expiration of a timeout T out, it will keep transmitting the data paket until a maximum transmission number N max is reahed. The timeout must be adjusted onsidering the long propagation and transmission delays, i.e., T out p(t + Tg ij +2 d ij / + q(t + Tg. Note that every time a transmission fails, sender i inreases the transmission power to improve the probability that the paket is suessfully reeived. V. POWER AND OFDM PARAMETERS SELF-ASSIGNMENT PROBLEM We onsider a shallow water aousti hannel, whih is severely affeted by multipath [1]. In suh environment, the signal fading an be modeled by a Rayleigh model. It is worth noting that suh model (Rayleigh model aounts for the worst-ase senario. The orresponding transmission loss TL ij that an aousti signal entered at frequeny f [Hz] experienes between nodes i and j ; d ij [m] distant, is TL ij = d ij 10 [α(f dij+a]/10 ρ 2 (1 where α (f represents the medium absorption oeffiient, A [5, 10] db is the so-alled transmission anomaly, whih aounts for the degradation of the aousti intensity aused by multiple path propagation, refration, diffration, and sattering of sound and ρ has a unit mean Rayleigh umulative distribution D ρ (ρ =1 exp ( ρ 2 /4. ρ is a normalized random variable that represents the power gain of the fading. Sine ρ is random, the reeived signal is also random. Hene, orret reeption of a signal an be guaranteed only on a probabilisti basis. In our work, we require that Pr{P rij τ} δ l for reliable reeption, where P rij is the power of the reeived signal, τ is a predefined power threshold, and δ l is the required link reliability. Given that the transmission power is, the orresponding reeived energy P rij is P rij = TL ij ρ 2 (2 The link-reliability requirement an be expressed as δ l Pr{P rij τ} Ptij Pr{ρ τ TL ij } 1 exp( Ptij 4τ TL ij 4τ d ij 10 [α(f dij+a]/10 log(1 δ l (4 Reall that the objetive of UW-OFDMAC protool is twofold: first minimizing energy onsumption and seond guaranteeing high bandwidth effiieny. In this work, we aim at minimizing the energy onsumption per symbol ( (T + Tg ij. To evaluate the bandwidth effiieny, let us onsider the symbol rate I R ij relative to transmitter i and reeiver j, R ij = i (T +Tg ij symbol per seond, and user i bandwidth is defined as.the resulting bandwidth effiieny is Rij =( (T + Tg ij 1. Now, in order to ombine both of our objetives into a ( single objetive funtion, we propose to minimize ( 1 Δfij P max (T + Tg ij Δf (T + Tg ij, where P max and Δf are only normalizing onstants. Consequently, our optimization problem an be stated as follows: given d ij,δ l,τ,v ij,,p max,t,,f i0 Find Δfij,P ( tij,tg ij ( 1 Δfij min (T + Tg ij (T + Tg ij,t g ij, P max Δf subjet to (5 Tg ij d ij (6 ( 1 max T, f ikv ij 4τ dij 10[α(f ik d ij +A]/10 log(1 δ l (3 (7 P max (8 f ik = f i0 + k (9 The first onstraint (6 states that the guard interval must be at least longer than the propagation delay in order to avoid ISI. The seond onstraint (7 states that the sub-arrier spaing should be i larger than the Doppler shift of the onsidered users i and j in order to avoid ICI and ii larger than 1 T in order to guarantee orthogonality. The Doppler shift of users i and j depends on v ij denoting the veloity of the transmitter i relative to the reeiver j in meters per seond. Preisely, the Doppler Shift that an aousti signal entered at frequeny f [Hz] experienes between nodes i and j with relative veloity v ij is f vij. Finally, the third onstraint (8 guarantees a reliable reeption of a data paket.

4 This full text paper was peer reviewed at the diretion of IEEE Communiations Soiety subjet matter experts for publiation in the IEEE ICC 2011 proeedings While our problem formulation may seem a fairly omplex optimization problem, it admits a low omplexity optimal solution. To find it, we rely on a major property of the objetive funtion, namely f (,Tg ij, = Ptij P max (T + Tg ij ( 1 Δfij Δf (T + Tg ij is monotonially and independently inreasing with,tg ij and. Hene, we are interested in taking the minimum values for,tg ij and in order to minimize f (,Tg ij,. Regarding the first onstraint, it is lear that the minimum value of Tg ij is dij. Hene, we get Tgij = d ij Aording to the seond onstraint (10 f ikv ij (11 ( 1 k vij f i0v ij (12 In this ase, the most rigorous inequality ( is ahieved when k = v. In other words, if 1 k ij f i0v ij is satisfied for k = then it will be automatially satisfied for all the remaining k s values. Therefore v ij (f i0 + (13 ( 1 Δfij = max T, v ij (f i0 + (14 Finally, regarding the third onstraint, we have 4τ d ij 10 [α(f ik d ij+a]/10 log(1 δ l (15 10 [α(f ik d ij+a]/10 4τ d ij log(1 δ l (16 In ( fresh water, α (f ik = fik 2 C fresh, Cfresh = , hene, we get P 10 log 10 ( tij 4τ d ij log(1 δ l A 1 (f i0 + k 2 (17 C fresh d ij Here again, the most rigorous inequality is ahieved when k =, hene we get Ptij = 4τ d ij 10 [(fi0+ 2 C fresh d ij+a]/10 log(1 δ l (18 VI. PERFORMANCE EVALUATION In this setion, we evaluate the effiieny of our proposed UW-OFDMAC protool for 3D shallow water arhiteture. To ahieve this, we use the lassial OFDMA protool as baseline to whih the UW-OFDMAC improvements an be ompared. By lassial OFDMA protool we mean the basi version with no dynami OFDMA parameters (Δf,Tg assignment. In other words, the OFDMA parameters will be fixed during the network setup. Note that, sine the objetive of our protool is to provide a ollision free aess for under-water sensor nodes, we also tune, for fairness purposes, the basi OFDMA protool to guarantee no ollision for sending sensors. We onsider 20 sensors initially deployed aording to fig. 1, in a 3D shallow water with volume of m 3, Fig. 1. Underwater sensor nodes deployement. whih may represent a small harbor. The surfae station whih represents the final destination for eah sent paket is entered on the surfae of the underwater volume. The routing shema is depited by the arrows in fig. 1. We model node s motion by a sinusoidal speed funtion with random amplitude varying between 0-5m/s. We set the maximum transmission power P max to 10W, the data paket size to 250 Bytes, the ACK and notifiation paket size to 10 Bytes, the available aousti spetrum to 50 khz, and the transmission anomalies aused by multipath in shallow water to 5 db. Moreover, all deployed sensors are soures, with paket inter-arrival time equal to 20 s. Sine the handled traffi inreases with nodes depth (Z(m, the nodes sub-hannel assignment shema provides inreasing spetrum for eah depth level. Aordingly, nodes at the first depth level (Z = 10m will be assigned onseutive sub-hannels of 1KHz, nodes at the seond depth level (Z = 30m will have 2KHz spetrum. Finally, eah node in the third and fourth depth levels will aquire spetrum of 3 and 4 KHz respetively. Fig. 2 ompares the energy onsumption per sensor node for both protools UW-OFDMAC and basi OFDMA. The most striking result is that UW-OFDMAC is ahieving a onsiderable energy savings (up to 96.7% for node 18 ompared with the basi OFDMA protool, whih justify the usefulness of our dynami OFDMA parameters assignment sheme. As explained above, the basi OFDMA protool has fixed parameters sine nodes deployment. Dealing with the energy onsumption (see eq. 18, the distane separating any two ommuniating nodes is tuned to the maximal possible distane between any sensor and its orresponding next hop. As suh, reliable transmission is guaranteed even with the basi OFDMA protool for eah pair of nodes. Now, let s fous on the energy onsumption ahieved by the UW-OFDMAC protool. Aording to fig. 3, the energy onsumption per node for UW-OFDMAC mainly depends on the maximal distane between a node and its orresponding next hop. In our simulations, before eah new transmission, d ij is hosen to be the maximal distane between i and j in the time domain. Reall that eah node in our simulations has a variable trajetory with time. Consequently, a rigorous appliation of our UW-OFDMAC protool requires OFDMA parameters omputing for eah transmitted byte whih highly inreases protool omplexity. Instead, we propose using the maximal possible parameters during a paket transmission. For instane, to ompute the optimal transmission power we need to know the urrent distane between a node and its next

5 This full text paper was peer reviewed at the diretion of IEEE Communiations Soiety subjet matter experts for publiation in the IEEE ICC 2011 proeedings Fig. 2. Energy Consumption per sensor node. Fig. 4. Bandwidth Effiieny per node. Fig. 3. Next hop distane vs. Next hop veloity. hop. Knowing that the speed period is lower than the paket transmission time, allows us to hoose the maximal distane in order to ensure ollision free aess during paket transmission. To summarize, the basi OFDMA protool will be tuned to the maximal possible parameters between any nodes and its orresponding next hop. However, UW-OFDMAC hooses the maximal parameters between a given node and its orresponding next hop. Fig. 4 ompares the bandwidth effiieny per node for both protools UW-OFDMAC and basi OFDMA. Here again, UW- OFDMAC learly outperforms the basi OFDMA. Reall that the bandwidth effiieny (( (T + Tg ij 1 depends on and Tg ij whih in turn depends on nodes relative veloity and the separating distane respetively. Aording to fig. 3, the bandwidth effiieny is inversely proportional to the distane with a fair modulation by nodes relative veloity. REFERENCES [1] I. F. Akyildiz, D. Pompili, and T. Melodia, Underwater Aousti Sensor Networks: Researh Challenges, Ad Ho Networks (Elsevier, vol. 3, no. 3, pp , May [2] E. Sozer, M. Stojanovi, and J. Proakis, Underwater Aousti Networks, IEEE Journal of Oeani Engineering, vol. 25, no. 1, pp , Jan [3] J. Proakis, E. Sozer, J. Rie, and M. Stojanovi, Shallow Water Aousti Networks, IEEE Communiations Magazine, pp , Nov [4] M. Stojanovi, Aousti (Underwater Communiations, in Enylopedia of Teleommuniations, J. G. Proakis, Ed. John Wiley and Sons, [5] R. F. Ormondroyd, A Robust Underwater Aousti Communiation System using OFDM-MIMO, OCEANS 2007 Europe, pp. 1 6, Aberdeen, Sotland, June [6] B. C. Kim and I. T. Lu, Parameter Study of OFDM Underwater Communiations System, OCEANS 2000, MTS/IEEE Conferene and Exhibition, Vol. 2, pp.: , Providene, Rhode Island, USA, Sept [7] S. Kumar, V. S. Raghavanb, and J. Deng, Medium Aess Control Protools for Ad Ho Wireless Networks: A Survey, Ad Ho Networks (Elsevier, vol. 4, no. 3, pp , May [8] [11] K. Kredo II and P. Mohapatra, Medium Aess Control in Wireless Sensor Networks, Computer Networks (Elsevier, vol. 51, no. 4, pp , Mar [9] D. Pompili, T. Melodia, and I. F. Akyildiz, A CDMA-based Medium Aess Control Protool for Underwater Aousti Sensor Networks, IEEE Transations on Wireless Communiations, vol. 8, no. 4, pp , April [10] M. Hayajneh, I. Khalil, Y. Gadallah, An OFDMA-based MAC protool for under water aousti wireless sensor networks, Proeedings of the 2009 International Conferene on Wireless Communiations and Mobile Computing, IWCMC 09, June 2009, Leipzig, Germany. VII. CONCLUSION Coneiving effiient MAC protools is a ritial issue in underwater aousti networks. In this paper, UW-OFDMAC, a distributed MAC protool for underwater aousti sensor networks, was proposed. It is an OFDMA based sheme that inorporates a distributed algorithm to set the optimal transmit power and OFDMA parameters. We have demonstrated that UW-OFDMAC manages to simultaneously ahieve low energy onsumption and high bandwidth effiieny in shallow water ommuniations, known to be heavily affeted by multipath fading.