Home & Garden

5 pages
5 views

Millimeter Wave WPAN: Cross-Layer Modeling and Multi-Hop Architecture

of 5
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Share
Description
Millimeter Wave WPAN: Cross-Layer Modeling and Multi-Hop Architecture
Transcript
  Millimeter Wave WPAN: Cross-Layer Modelingand Multihop Architecture Sumit Singh ∗ , Federico Ziliotto † , Upamanyu Madhow ∗ , Elizabeth M. Belding ∗ and Mark J. W. Rodwell ∗∗ University of California, Santa Barbara † University of Padova, Padova, Italy  Abstract —The 7 GHz of unlicensed spectrum in the 60GHz band offers the potential for multiGigabit indoor wirelesspersonal area networking (WPAN). With recent advances in thespeed of silicon (CMOS and SiGe) processes, low-cost transceiverrealizations in this “millimeter (mm) wave” band are withinreach. However, mm wave communication links are more fragilethan those at lower frequencies (e.g., 2.4 or 5 GHz) becauseof larger propagation losses and reduced diffraction aroundobstacles. On the other hand, directional antennas that providedirectivity gains and reduction in delay spread are far easier toimplement at mm-scale wavelengths. In this paper, we present across-layer modeling methodology and a novel multihop mediumaccess control (MAC) architecture for efficient utilization of 60 GHz spectrum, taking into account the preceding physicalcharacteristics. We propose an in-room WPAN architecture inwhich every link is constrained to be directional, for improvedpower efficiency (due to directivity gains) and simplicity of implementation (due to reduced delay spread). We develop anelementary diffraction-based model to determine network linkconnectivity, and define a multihop MAC protocol that accountsfor directional transmission/reception, procedures for topologydiscovery and recovery from link blockages. I. I NTRODUCTION The 60 GHz band has been allocated worldwide for shortrange wireless communications because high atmosphericpath loss due to oxygen absorption renders it unsuitablefor long distance communications [1], [2]. This abundantunlicensed spectrum has the potential to enable numerousindoor wireless applications that require large bandwidth suchas streaming content download (for High Definition Television(HDTV), video on demand, home theater, etc.); high speedInternet access and wireless gigabit Ethernet for laptops anddesktops [3]. These applications cannot be supported overexisting home networking solutions (IEEE 802.11 a/b/g)because the required data rates far exceed the capabilitiesof these networks. As a result, 60 GHz communication isattracting significant interest from both industry and academia,leading to active research and standardization efforts for thistechnology [4], [5], [6]. The IEEE 802.15 WPAN MillimeterWave Alternative PHY Task Group 3c formed in March 2005is working towards standardizing the physical (PHY) layerfor WPANs operating in the 60 GHz band [3].To leverage the potential of 60 GHz communication, wepresent a cross-layer modeling methodology and a novelmultihop MAC architecture for robust, multiGigabit, in-roomWPANs using 60 GHz mm wave spectrum. The successful har-nessing of 60 GHz spectrum for multiGigabit indoor WPANsrequires cross-layer design based on an understanding, at leastat a coarse level, of the unique physical layer properties of mm wave communication. We enumerate some relevant designconsiderations below, and contrast 60 GHz communicationwith that of the familiar IEEE 802.11a 5 GHz microwave band: •  Since free space propagation loss scales up as the square of the carrier frequency, the propagation loss for 60 GHz is morethan 20 dB worse than that at 5 GHz. •  Directional antennas are far easier to implement at 60GHz than at 5 GHz because of the smaller wavelengths.Directivity gains of 10-20 dB at each end are therefore easyto obtain at 60 GHz, which more than compensates for thehigher propagation loss. In addition, directional transmissionand reception simplifies the transceiver design by significantlyreducing the delay spread. •  Due to the smaller wavelength, the capability to diffractaround obstacles is far less for 60 GHz than for 5 GHz; e.g.,for directional links, a human in the line of sight (LOS) pathbetween the transmitter and receiver can attenuate the signalby 20-30 dB, effectively resulting in link outage. •  Due to the high attenuation of mm waves by obstacles, therange for an indoor mm wave network is of the order of 10meters, sufficient for an in-room WPAN. The transmit powerrequired for such ranges is small enough to be realizable withICs using low-cost CMOS and SiGe semiconductor processes. •  Due to the high bandwidth of operation, 60 GHz transceiversmust use high sampling rates. Since high-resolution, high-speed analog-to-digital converters are both expensive andpower-hungry, it is important to simplify the digital signalprocessing at the physical layer. Directional transmission andreception facilitates this by reducing the delay spread, andhence the intersymbol interference.Motivated by the preceding considerations, we proposethe following architecture for an in-room 60 GHz WPAN.Each node in the network is equipped with an electronicallysteerable directional antenna, and transmitters and receiverscan steer beams towards each other (this implies, for example,that a given node cannot be expected to hear transmissionsintended for other nodes). The network employs multihoprouting based on directional, LOS links. Each link operates ata fixed nominal data rate (e.g., 2 Gbps) when the LOS path isavailable. If the LOS path is blocked, we simply route aroundthe link. This design choice improves power efficiency. Alink whose LOS path is blocked might still be able to operateat a lower rate, exploiting reflections that reach the receiver.   0743-166X/07/$25.00 ©2007 IEEE This full text paper was peer reviewed at the direction of IEEE Communications Society subject matter experts for publication in the IEEE INFOCOM 2007 proceedings.  However, if the received power is, say, 10 dB lower, then wemust reduce the bit rate by a factor of 10 in order to maintainthe same reliability, when operating in power-efficient mode.In contrast, replacing the blocked path by two links in amultihop architecture only reduces the throughput by afactor of two. Our central thesis is that, assuming that thereare enough spatially dispersed nodes (including relays, if necessary), multihop networking with directional LOS linksprovides the best of both worlds: high power efficiency androbust connectivity in the face of stationary and movingobstacles typical of living room and office environments.Our work is motivated by recent advances in mm wavecircuit design [4], [5] that indicate low-cost IC realizations of mm wave nodes should be available in the near future. Thispromised cost reduction, together with the fact that directionalantennas can be realized in small form factors, motivates ournetwork architecture featuring directional links and relays.Millimeter wave propagation measurement and modelinghave received extensive attention over the past decade. Mea-surement campaigns in indoor environments include [7], [8],[9], [10]. Many deterministic and statistical mm wave prop-agation models have been proposed based on channel mea-surement studies [10], [11]. However, most of these focus onomnidirectional transmission (and possibly directional recep-tion). The benefits of base station diversity in reducing link blockage for omnidirectional transmission is analyzed in [12]for a simple model of an office environment. The reduction of multipath for directional links is well known [9], [12], as is thesusceptibility of directional mm wave links to blockage dueto their weak diffraction characteristics [1], [9]. To the bestof our knowledge, there is no significant prior published work on the design of mm wave WPANs with directional links.II. P HYSICAL  L AYER  M ODEL We first give an example link budget for a LOS 60 GHz link,to give a feel for the feasibility of WPANs with directionalLOS links. We then abstract away from detailed design choicesto focus on the key bottleneck for mm wave communication:blockage by obstacles. Sample link budget:  The directivity of an antenna is the ratioof the maximum power density (watts/  m 2 ) to its average valueover a sphere, and can be approximated as [13]: D  = 40000 θ oHP  φ oHP  where  θ oHP   and  φ oHP   are the horizontal and verticalbeamwidths, respectively. For a WPAN application, we mightdesign an antenna element to have a horizontal beamwidthof   120 o and a vertical beamwidth of   60 o , which allows arough placement of nodes in order to ensure LOS to one ortwo neighbors. The directivity for such an element, whichcan be realized as a pattern of metal on circuit board, is5.55 (or 7.4 dB). If we put four such elements to form asteerable antenna array, we can get a directivity of 22 (or13.4 dB). Now, assuming an antenna directivity of 10 dB ateach end, we do a link budget for a QPSK system operatingat 2 Gbps. For a receiver noise figure of 6 dB, bit error rateof   10 − 9 , excess bandwidth of 33%, and assuming free spacepropagation, we obtain that the required transmit power fora nominal range of 10 meters is 36 mW, including a 10 dBlink margin. When split between four antenna elements, thistransmit power corresponds to 9 mW of power per antennaelement. RF front ends for obtaining these power levelsare realizable with CMOS or SiGe processes, indicating thefeasibility of low-cost, high-volume production of the kindsof WPAN nodes on which our architecture is based. Link outage - a worst-case abstraction:  We neglect thecontribution from the reflected signals to the received signalpower. This is because our goal is to show that robustconnectivity can be obtained using a multihop architecture,even in a worst-case scenario considering energy only fromthe LOS components. Furthermore, narrow beam directionalantennas along the LOS direction substantially reduce thecontribution of reflected multipath components [9], [12], [14].We make the following simplifying assumptions in mod-eling obstacles: 1) We assume that the attenuation due to anobstacle in the LOS path is so high that the energy of the signalpropagating through the obstacle is negligible. In other words,we only consider obstacles that can cause a significant atten-uation to a signal propagating through them. For mm waves,most of the common obstructions in indoor environments,such as human beings, thick walls and furniture, fall in thiscategory. Thus, the link gain is only due to diffraction aroundthe obstacle. 2) The human body is approximated as a perfectconducting cylinder, whose projection perpendicular to theplane of propagation is considered for diffraction calculations.Other obstacles are approximated in a similar manner.If the diffraction loss due to obstacles exceeds 10 dB fora link, then it is considered to be in outage. This model ispessimistic because link budgets are determined based on amaximum range of operation (10 meters in our case). Byabstracting away the dependence of connectivity on range, weobtain a worst-case network connectivity model that servesto stress-test our proposed multihop architecture. Diffraction due to obstacles:  Diffraction of electromagneticwaves can be explained by a fundamental principle from phys-ical optics: the Huygens’ principle. Reference [15] provides adetailed analysis of the phenomenon of diffraction on the basisof this principle and also the geometrical theory of diffraction.To evaluate the effect of obstacles in terms of power loss,we define diffraction loss as  g diff   =  E E  0 , where  E   is theelectric field at the point of observation with diffraction effectsand  E  0  is the electric field at the same point in an unobstructedenvironment. We consider the access point (AP) as the sourceand the wireless terminals (WTs) as the observation points.The electric field  E   at the WTs can be calculated using thediffraction analysis based on the Huygens’ principle.Consider the scenario illustrated in Fig. 1 where there aretwo obstacles between an AP and a WT. Our goal is toevaluate the diffraction loss because of these obstacles. Wefirst consider an observation point  ( z 2 ,x 2 )  on the line  L 2 that is parallel to the  X   axis and passes through the second This full text paper was peer reviewed at the direction of IEEE Communications Society subject matter experts for publication in the IEEE INFOCOM 2007 proceedings.  Fig. 1. Multiple obstacles scenario. obstacle. From the analysis for the single obstacle case [15],we find that the diffraction loss at  ( z 2 ,x 2 )  is  g diff  ( x 2 ) . Nowall points on line  L 2  form new secondary wave sources forfurther diffraction loss because of the second obstacle on line L 2  (Huygens’ principle). Thus, the total diffraction loss at theWT can be evaluated as g diff  ( x r ) =    ∞−∞ g diff  ( x 2 ) T  ( x 2 ) e − jβ ( xr − x 2)22( zr − z 2) dx 2 ,  (1)where  ( z r ,x r )  is the location of the obstacle,  β   =  2 πλ  isthe phase constant for wavelength  λ , and function  T  ( x ) =1  for  x  ∈ { obstacle } , and  0  otherwise. Equation 1 can beviewed as the convolution of   f  diff  ( x )  and  e − jβ  x 22( zr − z 2) , where f  diff  ( x ) =  g diff  ( x ) T  ( x ) . Therefore, F{ g diff  ( x r ) }  =  F{ f  diff  ( x ) }F{ e − jβ  ( x )22( zr − z 2) } ,  (2)which can be computed using the Fast Fourier Transform(FFT) and Inverse Fast Fourier Transform (IFFT) algorithms.This analysis can be extended to the  m  obstacle case asfollows: F{ g mdiff  ( x r ) }  =  F{ g ( m − 1) diff   ( x ) T  m ( x ) }F{ e − jβ  ( x )22( zr − zm ) } , where  g ( m − 1) diff   ( x )  is the diffraction loss due to the  ( m  −  1) th obstacle, evaluated at  ( z m ,x ) . We can use this relation recur-sively starting from the nearest obstacle to the AP and movingtowards the next obstacles and then use the IFFT to obtain thefinal diffraction loss at the WT  ( z r ,x r ) .III. D IRECTIONAL  MAC D ESIGN The key idea behind our multihop relay directional MACframework is to utilize a mix of the conventional AP-basedsingle wireless hop MAC architecture for primary connectivityand resort to the multihop ad hoc mode with intermediatenodes acting as relays (though still controlled by the AP) toprevent drastic reduction of data rates or link outage whenthe LOS component to a WT is obstructed. Because of directional transmissions at all nodes, the conventional carriersensing solutions are not suited for mm wave WPANs. Discovery algorithm:  During the network initializationphase, the AP sends a Hello message and waits for responsefrom WTs in each sector. On receiving a Hello message, theWTs adjust their antenna beams to maximize the receivedpower from the AP and respond with the Hello Response Fig. 2. An example MAC message sequence over a superframe. message. The WTs in each sector employ a Slotted Aloha [16]contention scheme, with probabilities of transmission dictatedby the AP. After performing this discovery process in eachsector and having formed a network topology map (the identi-ties of WTs in the network and the appropriate antenna arrayconfigurations/directions required to reach them), the AP iter-atively designates each WT among the registered nodes to per-form the same discovery procedure. Every WT in the network sends its network topology map to the AP after it completes itsnetwork discovery process. The topology maps created duringthe initialization phase are useful for lost node discovery anddata transmission procedures, as described below. Normal mode of operation:  The AP polls all WTs in eachsector to check connectivity to each WT and to check whetherany WT has data to transmit. Each WT must respond withina fixed interval, i.e., Poll Inter Frame Space (PIFS), with adata packet or with a  connection live  poll response message,even if it does not have data to transmit. The dwell time ineach sector depends on the data transmission requirements of the WTs in that sector. A WT can continue to receive or senddata packets until a maximum allowed time duration, calledthe  transmission opportunity  (TXOP) duration. This allowsthe WTs to better utilize the available LOS connectivity andalso minimizes the control overhead associated with datapacket transmissions. If the AP sends a data packet to a WT,the WT acknowledges the successful packet reception eitherby piggybacking an ACK message on the next data packetthat it has for the AP or by sending a separate ACK message. Trailing control phase:  The trailing control phase is usedby the AP to allow new nodes to register and perform anetwork discovery procedure while the network is operational.During the trailing control phase, the AP can also verifyits own topology map or designate registered WTs to verifytheir network topology maps by sequentially sending Hellomessages to each WT. The trailing control phase is limitedto a maximum duration, which is higher than the averagesuccessful discovery phase time of a node. Because the regularnetwork topology verification procedure of the trailing controlphase occurs at a rate much faster than the dynamics of theindoor environments (human movements or change in roomsetup), the AP is aware of LOS connectivity of all the WTs andit can use the topology verification/discovery reports sent back by the WTs to choose a candidate relay node for a lost WT.The trailing control phase also allows the AP to take care of other control requirements such as  superframe  end signaling,where a superframe is defined as the time taken by the APto poll all the registered WTs in the network. The maximum This full text paper was peer reviewed at the direction of IEEE Communications Society subject matter experts for publication in the IEEE INFOCOM 2007 proceedings.  Parameter Value Human height range (1.5m - 2.1m)Random Waypoint model velocities (min,max) (0m/s,1m/s)Random Waypoint model pause time 10sFixed obstacle height range (1m - 1.4m)WT location height range (0.5m - 1.5m)AP location height 2mSimulation time 5minTABLE IE VALUATION  M ODEL  P ARAMETERS superframe duration is limited by the number of WTs in thenetwork, the TXOP duration, and the trailing control phase du-ration. The AP signals the end of a superframe after the trailingcontrol phase functions are finished. Fig. 2 illustrates an exam-ple of data transmission and control message sequence over asuperframe for a network comprised of an AP and five WTs. Lost node discovery:  If the AP does not receive a pollresponse from a registered WT within the PIFS interval, itconsiders the WT to be lost and intelligently chooses a WTamong the live WTs in the neighboring sectors (with expectedLOS connectivity to the lost WT as determined from theregular topology verification reports from the WTs) to act asa relay to the lost node. It commands the chosen relay WTto discover (i.e., check connectivity status with) the lost WTand report back within a stipulated time. The designated relayWT refers to its network topology map information to steerits antenna beam to the lost WT, and sends a Hello messageto the lost WT. If the lost WT is able to receive this message,it adjusts its antenna beam towards the designated relay WTand responds with a Hello Response message. Upon receivinga reply from the lost WT, the chosen relay WT reports to theAP the quality of the link (i.e., the received signal strength)between itself and the lost WT. Otherwise, after waitingfor a PIFS interval, it informs the AP of discovery failure.Depending on the response from the designated relay WT, theAP decides whether to choose another WT in the poll sequenceto repeat the lost node discovery procedure or to use thecurrent chosen WT as a relay for future data transfers until theLOS connectivity to the lost WT is restored. Upon a successfullost WT discovery, the AP adds the required data transfer timefor the lost WT to the relay WT’s sector dwell time. Once theobstruction is removed and the lost WT starts receiving directtransmissions from the AP, it responds to the AP’s poll mes-sage. The AP switches back to the normal mode of operationafter informing the relay WT to return to its previous state.IV. P ERFORMANCE  E VALUATION In this section, we describe our evaluation methodology andpresent performance results for our multihop relay directionalMAC scheme in comparison to the conventional single hopAP to WT communication MAC scheme defined for theinfrastructure mode of the IEEE 802.11 MAC. We call thesingle hop communication scheme as the baseline MAC.We have developed a MatLab tool to evaluate theperformance of our multihop relay MAC scheme by 01234560123456X axis [m]    Z  a  x   i  s   [  m   ] AP WT 1WT 2 WT 3 WT 4WT 5 WT 6WT 7 WT 8 Fig. 3. Living room scenario.   0 20 40 60 80 100 1200102030405060Time [s]    L  o  s  s   [   d   B   ] WT2 WT5 Fig. 4. AP to WT LOS link lossprofile. simulating different indoor environments with human beingsand other obstacles. This tool is based on our deterministicphysical radio propagation model (see Section II) which yieldslink connectivity between different network nodes given thelink margin. The inputs to our tool are parameters required tosimulate a WPAN in a specified 3-dimensional indoor envi-ronment: the room dimensions, the positions and dimensionsof stationary obstacles such as furniture, the number of humanbeings, and the placement of the AP and the WTs. We use theRandom Waypoint model for human movements in the room.Table I lists the default parameter values for our test scenarios.We define  connectivity consistency  as the percentage of timeout of the total operation period of the network when a WTis reachable from the AP, either through a direct LOS link orthrough a relay node. Thus, this metric characterizes the actualconnection state and data transfer capacity of the network. Wealso evaluate the expected aggregate network throughput andstudy its variation over time with respect to dynamics of theindoor environments such as obstacle movements.We consider a living room that has a WPAN formed byan HDTV, a surround sound system with speakers at roomcorners, and a desktop/printer; and has eight human beings,i.e., during a gathering at home (see Fig. 3). This models atypical environment where 60 GHz WPANs are expected tobe deployed. The room and obstacle dimensions and nodeplacements have been chosen as representative of real worldscenarios in which a large number of people can cause ahigh blockage probability for individual links. Note that WT1is placed higher than the other WTs (2.5m, compared to1.2m on average for the other WTs) such that it has a highprobability of a clear LOS connectivity to most of the WTsand the AP. Hence WT1 can act as a relay in case the directLOS connectivity from the AP to a WT is blocked.Fig. 4 plots the link loss variation for two specific WT links(WT2 and WT5) over a sample period of 120 seconds. Weobserve that there are heavy link losses because of the largenumber of human beings and stationary furniture obstacles.These obstacles result in intermittent connectivity to the af-fected WT if the underlying MAC completely relies on directsingle hop connectivity of the AP to WTs. Thus, the baselineMAC scheme cannot provide the required QoS or data rateguarantees to different WPAN applications. This full text paper was peer reviewed at the direction of IEEE Communications Society subject matter experts for publication in the IEEE INFOCOM 2007 proceedings.  01 0011 Single HopMultihop Relay 000000000000000000000000000000000000000000000000111111111111111111111111111111111111111111111111 0000000000000011111111111111 0000000000000000000000000000000000000011111111111111111111111111111111111111 0000000000011111111111 0000000000000000011111111111111111 00000000000000000000000000000000000000000000001111111111111111111111111111111111111111111111 00000000000000000000000000000000000000000000001111111111111111111111111111111111111111111111 00000000000000000000000000001111111111111111111111111111 0000000000000000000000011111111111111111111111 0000000000000000000000011111111111111111111111 00000000000000000000000000000000000000000000001111111111111111111111111111111111111111111111 0000000000000000000000011111111111111111111111 00000000000000000000000000000000000000000000001111111111111111111111111111111111111111111111 0000000000000000000000011111111111111111111111 0000000000000000000000011111111111111111111111 00000000000000000000000000000000000000000000001111111111111111111111111111111111111111111111  0% 20% 40% 60% 80% 100%WT8WT7WT6WT5WT4WT3WT2WT1    C  o  n  n  e  c   t   i  v   i   t  y  c  o  n  s   i  s   t  e  n  c  y   (   %   ) Fig. 5. AP − WT connectivity consistency.   0 20 40 60 80 100 1200.50.70.91.11.31.5Time [s]    A   g   g   r   e   g   a   t   e   t   h   r   o  u   g   h   p  u   t   [   G   b   p   s   ] Fig. 6. Aggregate network throughput. 0 20 40 60 80 100 1200123456Time [s]    W   T   s   o   n   m  u   l   t   i   h   o   p   c   o   n   n   e   c   t   i  v   i   t  y Fig. 7. Number of WTs connected viarelays. Fig. 5 compares the WT connectivity consistency for thebaseline MAC and the directional multihop relay MAC. Weobserve that, on average, connectivity consistency in the base-line case is significantly lower than the multihop relay MACscheme, which is able to maintain almost 100% network con-nectivity. Note that the availability of alternate routes in a mul-tihop architecture can be easily ensured by appropriate place-ment of relay nodes (e.g., high up on walls, or on the ceiling).On the other hand, the poor connectivity consistency of singlehop communication makes it unsuitable for WPAN applica-tions with stringent QoS requirements such as video streaming.Fig. 6 plots the expected aggregate network throughputvariation for the multihop relay MAC scheme. We observe thatthe aggregate throughput remains fairly consistent over time.We do not plot aggregate throughput for the single hop MACsince we have already seen its poor connectivity consistency.Fig. 7 depicts the number of WTs connected via multihoppaths at different sampling instances of the simulation. Asignificant number of WTs using multihop relays at any timeinstant shows the importance of multihop paths in maintainingcontinuous network connectivity.V. C ONCLUSIONS Our results illustrate the critical role of cross-layer design inexploiting the large unlicensed bandwidth available in the 60GHz band. The simple diffraction-based connectivity model isan effective tool for cross-layer design: it yields results thatconform to our intuition that directional LOS mm wave linksexperience relatively high levels of outage due to stationaryand moving obstacles. Despite this fragility of individual links,we show that the proposed multihop MAC architecture is suc-cessful in providing robust connectivity in typical “SuperbowlParty” settings. Thus, unlike infrastructure mode operation in2.4 GHz and 5 GHz WLANs, where WTs communicate withAPs over a single hop, we believe that multihop communica-tion, possibly with nodes explicitly designated as relays, mustplay a fundamental role in 60 GHz WPANs. Our future work therefore focuses on refining the proposed multihop designbased on more detailed modeling and simulation. At the phys-ical layer, detailed evaluation of packetized systems for spe-cific modulation formats (e.g., single carrier modulation andOFDM) is necessary. At the application layer, it is of interest toevaluate MAC performance in more detail using traffic modelsaimed specifically at some of the entertainment applicationsdriving the interest in high-speed WPANs, such as streamingcompressed and uncompressed HDTV, and large file transfers.A CKNOWLEDGMENT This work was supported by the National Science Foun-dation under grants ANI-0220118, ECS-0636621 and NSFCareer Award CNS-0347886.R EFERENCES[1] P. F. M. Smulders, “Exploiting the 60 GHz Band for Local WirelessMultimedia Access: Prospects and Future Directions,”  IEEE Commun. Mag. , vol. 40, no. 1, pp. 140–147, Jan. 2002.[2] F. Giannetti, M. Luise, and R. Reggiannini, “Mobile and Personal Com-munications in the 60 GHz Band: A Survey,”  Wireless Pers. Commun. ,vol. 10, no. 2, pp. 207–243, Jul. 1999.[3] (2004, March) IEEE 802.15 WPAN Millimeter Wave Alternative PHYTask Group 3c (TG3c). [Online]. Available: http://www.ieee802.org/15/ pub/TG3c.html[4] (2006) IBMs 60-GHz Page. [Online]. Available: http://domino.research.ibm.com/comm/research projects.nsf/pages/mmwave.%sixtygig.html[5] (2006) 60 GHz CMOS Radio Design at Berkeley Wireless ReaearchCenter. [Online]. Available: http://bwrc.eecs.berkeley.edu/Research/RF/ ogre project/ [6] (2006) WIGWAM - Wireless Gigabit with Advanced MultimediaSupport. [Online]. Available: http://www.wigwam-project.de/ [7] H. Xu, V. Kukshya, and T. S. Rappaport, “Spatial and Temporal Char-acteristics of 60 GHz Indoor Channels,”  IEEE J. Sel. Areas Commun. ,vol. 20, no. 3, pp. 620–630, Apr. 2002.[8] P. F. M. Smulders, “Broadband Wireless LANs: A Feasibility Study,”Ph.D. dissertation, Eindhoven University of Technology, The Nether-lands, 1995.[9] T. Manabe, K. Sato, H. Masuzawa, K. Taira, T. Ihara, Y. Kasashima, andK. Yamaki, “Effects of Antenna Directivity and Polarization on IndoorMultipath Propagation Characteristics at 60 GHz,”  IEEE J. Sel. AreasCommun. , vol. 14, no. 3, pp. 441–448, Apr. 1996.[10] J. Kunisch, E. Zollinger, J. Pamp, and A. Winkelmann, “MEDIAN60GHz Wideband Indoor Radio Channel Measurements and Model,”in  Proc. IEEE VTC’99 , Amsterdam, The Netherlands, Sep. 1999.[11] P. F. M. Smulders, “Deterministic Modelling of Indoor Radio Propaga-tion at 40-60 GHz,”  Wireless Pers. Commun. , vol. 1, no. 2, pp. 127–135,Jun. 1994.[12] K. Sato and T. Manabe, “Estimation of Propagation-Path Visibility forIndoor Wireless LAN Systems under Shadowing Condition by HumanBodies,” in  Proc. IEEE VTC’98 , Ottawa, Canada, May 1998.[13] J. D. Kraus,  Antennas for all Applications.  New York: McGraw-Hill,Inc., 2002.[14] M. Williamson, G. E. Athanasiadou, and A. R. Nix, “Investigating theEffects of Antenna Directivity on Wireless Indoor Communication at60GHz,” in  Proc. IEEE PIMRC’97  , Helsinki, Finland, Sep. 1997.[15] J. D. Kraus,  Electromagnetics . New York: McGraw-Hill, Inc., 1991.[16] N. Abramson, “The Throughput of Packet Broadcasting Channels,”  IEEE Trans. Commun. , vol. 25, no. 1, pp. 117–128, Jan. 1977. This full text paper was peer reviewed at the direction of IEEE Communications Society subject matter experts for publication in the IEEE INFOCOM 2007 proceedings.
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks