cooperation in wireless networks principle and applications.rar_spring_ios_其他

<html xmlns="http://www.w3.org/1999/xhtml"> <head> <meta charset="utf-8"> <meta name="generator" content="pdf2htmlEX"> <meta http-equiv="X-UA-Compatible" content="IE=edge,chrome=1"> <link rel="stylesheet" href="https://static.pudn.com/base/css/base.min.css"> <link rel="stylesheet" href="https://static.pudn.com/base/css/fancy.min.css"> <link rel="stylesheet" href="https://static.pudn.com/prod/directory_preview_static/62546b02b744eb386faa74c2/raw.css"> <script src="https://static.pudn.com/base/js/compatibility.min.js"></script> <script src="https://static.pudn.com/base/js/pdf2htmlEX.min.js"></script> <script> try{ pdf2htmlEX.defaultViewer = new pdf2htmlEX.Viewer({}); }catch(e){} </script> <title></title> </head> <body> <div id="sidebar" style="display: none"> <div id="outline"> </div> </div> <div id="pf1" class="pf w0 h0" data-page-no="1"><div class="pc pc1 w0 h0"><img class="bi x0 y0 w1 h1" alt="" src="https://static.pudn.com/prod/directory_preview_static/62546b02b744eb386faa74c2/bg1.jpg"><div class="t m0 x1 h2 y1 ff1 fs0 fc0 sc0 ls0 ws0">Chapter 9</div><div class="t m0 x1 h3 y2 ff2 fs0 fc0 sc0 ls1 ws0">COGNITIVE RADIO ARCHITECTURE</div><div class="t m0 x1 h4 y3 ff3 fs0 fc0 sc0 ls0 ws0">Organizing Computational Intelligence for Peer and</div><div class="t m0 x1 h4 y4 ff3 fs0 fc0 sc0 ls0 ws0">Network Collaboration</div><div class="t m0 x1 h5 y5 ff1 fs1 fc0 sc0 ls2 ws0">Joseph Mitola III</div><div class="t m0 x1 h6 y6 ff3 fs2 fc0 sc0 ls0 ws0">The Mitre Corporation</div><div class="t m0 x2 h7 y7 ff4 fs3 fc0 sc0 ls0 ws0">∗</div><div class="t m0 x1 h8 y8 ff5 fs3 fc0 sc0 ls0 ws0">jmitola@mitre.org</div><div class="t m0 x1 h9 y9 ff2 fs2 fc0 sc0 ls0 ws0">Abstract: This chapter explores in detail the concept of cognitive radio architecture in cen-</div><div class="t m0 x3 h9 ya ff1 fs2 fc0 sc0 ls0 ws0">tralized and ad hoc networks. Such architecture is paramount for establishing</div><div class="t m0 x3 h9 yb ff1 fs2 fc0 sc0 ls0 ws0">a cooperating platform of environment-aware nodes. The chief constituent ele-</div><div class="t m0 x3 h9 yc ff1 fs2 fc0 sc0 ls3 ws0">ments of the cognitive radio architecture include functions implemented through</div><div class="t m0 x3 h9 yd ff1 fs2 fc0 sc0 ls0 ws0">components using a set of design rules. The synergy between software deﬁned</div><div class="t m0 x3 h9 ye ff1 fs2 fc0 sc0 ls0 ws0">radio and cognitive principles is fully exploited by assuming that the cognitive</div><div class="t m0 x3 h9 yf ff1 fs2 fc0 sc0 ls3 ws0">radio architecture is based on a software deﬁned radio endowed with a perceptive</div><div class="t m0 x3 h9 y10 ff1 fs2 fc0 sc0 ls0 ws0">system capable of sensing the radio environment, in particular the use of the ra-</div><div class="t m0 x3 h9 y11 ff1 fs2 fc0 sc0 ls3 ws0">dio spectrum. This capability is complemented with an intelligent system able to</div><div class="t m0 x3 h9 y12 ff1 fs2 fc0 sc0 ls3 ws0">understand the radio environment and, based on the user’s needs or expectations,</div><div class="t m0 x3 h9 y13 ff1 fs2 fc0 sc0 ls0 ws0">dynamically reconﬁgure the system taking into account spectrum usage. The so</div><div class="t m0 x3 h9 y14 ff1 fs2 fc0 sc0 ls3 ws0">called cognition cycle is an architecture subsystem including an inference hier-</div><div class="t m0 x3 h9 y15 ff1 fs2 fc0 sc0 ls4 ws0">archy, the temporal organization and ﬂow of inferences and control states. The</div><div class="t m0 x3 h9 y16 ff1 fs2 fc0 sc0 ls0 ws0">cycle continually observes the environment, orients itself, creates plans, decides,</div><div class="t m0 x3 h9 y17 ff1 fs2 fc0 sc0 ls0 ws0">and then acts. The phases of inference from observation to action show the ﬂow</div><div class="t m0 x3 h9 y18 ff1 fs2 fc0 sc0 ls3 ws0">of inference, a top-down view of how cognition is implemented algorithmically.</div><div class="t m0 x3 h9 y19 ff1 fs2 fc0 sc0 ls5 ws0">The inference hierarchy is the part of the algorithm architecture that organizes</div><div class="t m0 x3 h9 y1a ff1 fs2 fc0 sc0 ls0 ws0">the data structures. The cognition functions are implemented via cognition el-</div><div class="t m0 x3 h9 y1b ff1 fs2 fc0 sc0 ls0 ws0">ements consisting of data structures, processes and ﬂows. These include data</div><div class="t m0 x3 h9 y1c ff1 fs2 fc0 sc0 ls3 ws0">structures and related processing elements may be modeled as topological maps</div><div class="t m0 x3 h9 y1d ff1 fs2 fc0 sc0 ls0 ws0">over abstract domains. The processing elements of the architecture, modeled</div><div class="t m0 x3 h9 y1e ff1 fs2 fc0 sc0 ls3 ws0">topological maps like input, best-match and other maps, are described in detail in</div><div class="t m0 x1 ha y1f ff6 fs4 fc0 sc0 ls0 ws0">∗</div><div class="t m0 x4 hb y20 ff1 fs3 fc0 sc0 ls0 ws0">The author’s afﬁliation with the MITRE Corporation is provided for identiﬁcation purposes only, and</div><div class="t m0 x1 hb y21 ff1 fs3 fc0 sc0 ls0 ws0">is not intended to convey or imply MITRE’s concurrence with, or support for, the positions, opinions or</div><div class="t m0 x1 hb y22 ff1 fs3 fc0 sc0 ls0 ws0">viewpoints expressed by the author.</div><div class="t m0 x5 h5 y23 ff1 fs1 fc0 sc0 ls2 ws0">243</div><div class="t m0 x1 hc y24 ff3 fs3 fc0 sc0 ls6 ws0">Frank H. P. Fitzek and Marcos D. Katz (eds.), Cooperation in Wireless Networks: Principles and</div><div class="t m0 x1 hb y25 ff3 fs3 fc0 sc0 ls0 ws0">Applications, 243–311.</div><div class="t m0 x6 hb y26 ff1 fs3 fc0 sc0 ls0 ws0">c</div><div class="t m0 x1 h7 y27 ff4 fs3 fc0 sc0 ls0 ws0"> 2006 Springer. Printed in the Netherlands.</div></div><div class="pi" data-data='{"ctm":[1.611850,0.000000,0.000000,1.611850,0.000000,0.000000]}'></div></div> </body> </html>

<div id="pf2" class="pf w0 h0" data-page-no="2"><div class="pc pc2 w0 h0"><img class="bi x0 y0 w1 h1" alt="" src="https://static.pudn.com/prod/directory_preview_static/62546b02b744eb386faa74c2/bg2.jpg"><div class="t m0 x7 h5 y28 ff1 fs1 fc0 sc0 ls2 ws0">244 Cooperation in Wireless Networks</div><div class="t m0 x8 h9 y29 ff1 fs2 fc0 sc0 ls0 ws0">this chapter. As software deﬁned radio is the fundamental platform for cognitive</div><div class="t m0 x8 h9 y2a ff1 fs2 fc0 sc0 ls0 ws0">radio, the former is reviewed also in this chapter, including overviews of software</div><div class="t m0 x8 h9 y2b ff1 fs2 fc0 sc0 ls0 ws0">radio architecture (SRA) and software communications architecture (SCA), the</div><div class="t m0 x8 h9 y2c ff1 fs2 fc0 sc0 ls3 ws0">transition and evolution towards cognitive radio architecture, and other relevant</div><div class="t m0 x8 h9 y2d ff1 fs2 fc0 sc0 ls3 ws0">technical aspects.</div><div class="t m0 x7 hd y2e ff2 fs6 fc0 sc0 ls9 ws0">1. Introduction</div><div class="t m0 x9 he y2f ff1 fs7 fc0 sc0 ls0 ws0">The Cognitive Radio Architecture (CRA) frames the Cognitive Wireless</div><div class="t m0 x7 he y30 ff1 fs7 fc0 sc0 ls0 ws0">Networks (CWNs) for cooperation among computationally intelligent nodes</div><div class="t m0 x7 he y31 ff1 fs7 fc0 sc0 ls0 ws0">whether orchestrated by a host CWN or forming dynamic peer CWN on the ﬂy.</div><div class="t m0 x7 he y32 ff1 fs7 fc0 sc0 ls0 ws0">This chapter contributes to cooperative networks by examining the functions,</div><div class="t m0 x7 he y33 ff1 fs7 fc0 sc0 ls0 ws0">components, and design rules -the architecture - that enhances software-deﬁned</div><div class="t m0 x7 he y34 ff1 fs7 fc0 sc0 ls0 ws0">radio SDR to cognition. One critical differentiator between SDR and cognitive</div><div class="t m0 x7 he y35 ff1 fs7 fc0 sc0 ls0 ws0">radio is the ability to learn from the RF environment and the user. For this</div><div class="t m0 x7 he y36 ff1 fs7 fc0 sc0 ls0 ws0">learning to occur without burdening the user with “dumb questions” the cogni-</div><div class="t m0 x7 he y37 ff1 fs7 fc0 sc0 ls0 ws0">tive radio needs perception systems by which to perceive just those aspects of</div><div class="t m0 x7 he y38 ff1 fs7 fc0 sc0 ls0 ws0">the environment from RF and user viewpoints that enables the radio to better</div><div class="t m0 x7 he y39 ff1 fs7 fc0 sc0 ls0 ws0">support the quality of information (QoI) needed by the user.</div><div class="t m0 x7 hd y3a ff2 fs6 fc0 sc0 ls0 ws0">Cognitive Radios Know Radio Like TellMe</div><div class="t m0 xa hf y3b ff2 fs8 fc0 sc0 ls0 ws0">R</div><div class="t m0 xb hd y3a ff7 fs9 fc0 sc0 ls0 ws0"> Knows 800</div><div class="t m0 x7 hd y3c ff2 fs6 fc0 sc0 ls0 ws0">Numbers</div><div class="t m0 x9 he y3d ff1 fs7 fc0 sc0 ls0 ws0">When you dial 1-800-555-1212, an algorithm says “Toll Free Directory As-</div><div class="t m0 x7 he y3e ff1 fs7 fc0 sc0 ls0 ws0">sistance powered by TellMe</div><div class="t m0 xc h10 y3f ff1 fs8 fc0 sc0 ls0 ws0">R</div><div class="t m0 xd h11 y40 ff8 fs7 fc0 sc0 ls0 ws0"> [TellMe Networks, 2005]. Please say the name</div><div class="t m0 x7 he y41 ff1 fs7 fc0 sc0 ls0 ws0">of the listing you want.” If you mumble it says, “OK, United Airlines. If that</div><div class="t m0 x7 he y42 ff1 fs7 fc0 sc0 ls0 ws0">is not what you wanted press 9, otherwise wait while I look up the number.”</div><div class="t m0 x7 he y43 ff1 fs7 fc0 sc0 ls0 ws0">Ninety-nine point nine, nine percent of the time TellMe</div><div class="t m0 xe h10 y44 ff1 fs8 fc0 sc0 ls0 ws0">R</div><div class="t m0 xa h11 y45 ff8 fs7 fc0 sc0 ls0 ws0"> gets it right, replac-</div><div class="t m0 x7 he y46 ff1 fs7 fc0 sc0 lsb ws0">ing thousands of directory assistance operators of yore. TellMe</div><div class="t m0 xf h10 y47 ff1 fs8 fc0 sc0 ls0 ws0">R</div><div class="t m0 x10 h11 y48 ff8 fs7 fc0 sc0 ls0 ws0">, a speech-</div><div class="t m0 x7 he y49 ff1 fs7 fc0 sc0 ls0 ws0">understanding system, achieves a high degree of success by its focus on just</div><div class="t m0 x7 he y4a ff1 fs7 fc0 sc0 ls0 ws0">one task: ﬁnding a toll-free telephone number. Narrow task focus is the key to</div><div class="t m0 x7 he y4b ff1 fs7 fc0 sc0 lsb ws0">algorithm successes.</div><div class="t m0 x9 he y4c ff1 fs7 fc0 sc0 ls0 ws0">The Cognitive Radio Architecture (CRA) frames the Cognitive Wireless</div><div class="t m0 x7 he y4d ff1 fs7 fc0 sc0 lsb ws0">Networks (CWNs), offspring of TellMe</div><div class="t m0 x11 h10 y4e ff1 fs8 fc0 sc0 ls0 ws0">R</div><div class="t m0 x12 h11 y4f ff8 fs7 fc0 sc0 ls0 ws0">. CWNs are emerging as practical,</div><div class="t m0 x7 he y50 ff1 fs7 fc0 sc0 ls0 ws0">real-time, highly focused applications of computationally intelligent technol-</div><div class="t m0 x7 he y51 ff1 fs7 fc0 sc0 ls0 ws0">ogy. CWNs differ from the more general Artiﬁcial Intelligence (AI) based ser-</div><div class="t m0 x7 he y52 ff1 fs7 fc0 sc0 ls0 ws0">vices like intelligent agents, computer speech, and computer vision in degree</div><div class="t m0 x7 he y53 ff1 fs7 fc0 sc0 lsb ws0">of focus. Like TellMe</div><div class="t m0 x13 h10 y54 ff1 fs8 fc0 sc0 ls0 ws0">R</div><div class="t m0 x14 h11 y55 ff8 fs7 fc0 sc0 ls0 ws0">, CWNs focus on a very narrow tasks. Broader than</div><div class="t m0 x7 he y56 ff1 fs7 fc0 sc0 lsb ws0">TellMe</div><div class="t m0 x15 h10 y57 ff1 fs8 fc0 sc0 ls0 ws0">R</div><div class="t m0 x16 h11 y58 ff8 fs7 fc0 sc0 ls0 ws0">, the task is to create adapt radio-enabled information services for</div><div class="t m0 x7 he y59 ff1 fs7 fc0 sc0 ls0 ws0">to the speciﬁc needs of a speciﬁc user. TellMe</div><div class="t m0 x17 h10 y5a ff1 fs8 fc0 sc0 ls0 ws0">R</div><div class="t m0 x18 h11 y5b ff8 fs7 fc0 sc0 ls0 ws0">, a network service, requires</div><div class="t m0 x7 he y5c ff1 fs7 fc0 sc0 ls0 ws0">substantial network computing resources, teraﬂops of computing capability and</div><div class="t m0 x7 he y5d ff1 fs7 fc0 sc0 ls0 ws0">terabytes of memory to serve thousands of users at once. CWNs, on the other</div><div class="t m0 x7 he y5e ff1 fs7 fc0 sc0 ls0 ws0">hand, may start with a Cognitive Radio (CR) in your purse or on your belt, a</div><div class="t m0 x7 he y5f ff1 fs7 fc0 sc0 lsb ws0">cell phone on steroids, focused on the narrow task of creating from the myriad</div></div><div class="pi" data-data='{"ctm":[1.611850,0.000000,0.000000,1.611850,0.000000,0.000000]}'></div></div>

<div id="pf3" class="pf w0 h0" data-page-no="3"><div class="pc pc3 w0 h0"><img class="bi x0 y0 w1 h1" alt="" src="https://static.pudn.com/prod/directory_preview_static/62546b02b744eb386faa74c2/bg3.jpg"><div class="t m0 x1 h5 y28 ff3 fs5 fc0 sc0 ls8 ws0">Cognitive Radio Architecture 245</div><div class="t m0 x1 he y60 ff1 fs7 fc0 sc0 ls0 ws0">available wireless information networks and resources just what is needed by</div><div class="t m0 x1 he y61 ff1 fs7 fc0 sc0 lsb ws0">just one user, you. TellMe</div><div class="t m0 x19 h10 y62 ff1 fs8 fc0 sc0 ls0 ws0">R</div><div class="t m0 x1a h11 y63 ff8 fs7 fc0 sc0 ls0 ws0"> interacts with anybody, but each CR is self-aware</div><div class="t m0 x1 he y64 ff1 fs7 fc0 sc0 ls0 ws0">and owner-aware via sensory perception and Autonomous Machine Learning</div><div class="t m0 x1 he y65 ff1 fs7 fc0 sc0 ls0 ws0">(AML) technologies, earning the term “cognitive.” Each CR fanatically serves</div><div class="t m0 x1 he y66 ff1 fs7 fc0 sc0 ls0 ws0">the needs and protects the personal information of just one user, you, its owner.</div><div class="t m0 x1 he y67 ff1 fs7 fc0 sc0 ls0 ws0">Thus, via the CRA incorporates with perception and AML.</div><div class="t m0 x1b he y68 ff1 fs7 fc0 sc0 lsb ws0">TellMe</div><div class="t m0 x1c h10 y69 ff1 fs8 fc0 sc0 ls0 ws0">R</div><div class="t m0 x1d h11 y6a ff8 fs7 fc0 sc0 ls0 ws0"> is here and now, while CWNs are emerging in global wireless</div><div class="t m0 x1 he y6b ff1 fs7 fc0 sc0 lsb ws0">research centers and industry forums like the SDR Forum and Wireless World</div><div class="t m0 x1 he y6c ff1 fs7 fc0 sc0 ls0 ws0">Research Forum (WWRF). This chapter provides an overview of CRA sys-</div><div class="t m0 x1 he y6d ff1 fs7 fc0 sc0 ls0 ws0">tems architecture challenges and approaches, emphasizing CR as a technology</div><div class="t m0 x1 he y6e ff1 fs7 fc0 sc0 ls0 ws0">enabler for rapidly emerging commercial CWN services and generation-after-</div><div class="t m0 x1 he y6f ff1 fs7 fc0 sc0 ls0 ws0">next military communications, based on the foundation technologies computer</div><div class="t m0 x1 he y70 ff1 fs7 fc0 sc0 ls0 ws0">vision, computer speech, AML, and SDR.</div><div class="t m0 x1 hd y71 ff2 fs6 fc0 sc0 ls0 ws0">CRs See What You See to Discover RF, Needs, and Preferences</div><div class="t m0 x1b he y72 ff1 fs7 fc0 sc0 ls0 ws0">In 2002, GRACE (Graduate Robot Attending ConferencE) [Anne Watzman,</div><div class="t m0 x1 he y73 ff1 fs7 fc0 sc0 ls0 ws0">2002], Roombah’s big sister with a CRT for a face entered the International</div><div class="t m0 x1 he y74 ff1 fs7 fc0 sc0 ls0 ws0">Joint Conference on Artiﬁcial Intelligence (IJCAI); found the registration desk;</div><div class="t m0 x1 he y75 ff1 fs7 fc0 sc0 ls0 ws0">registered by talking to the receptionist; followed the signs that said “ROBOTS”</div><div class="t m0 x1 he y76 ff1 fs7 fc0 sc0 ls0 ws0">this way and “HUMANS” the other way; when called on gave a ﬁve-minute talk</div><div class="t m0 x1 he y77 ff1 fs7 fc0 sc0 lsb ws0">about herself; and then answered questions. She was the ﬁrst to complete the</div><div class="t m0 x1 he y78 ff1 fs7 fc0 sc0 ls0 ws0">mobile robot challenge ﬁrst articulated in the 1980’s. There were no joysticks</div><div class="t m0 x1 he y79 ff1 fs7 fc0 sc0 ls0 ws0">and no man behind the curtain: just a robot that can autonomously see, hear,</div><div class="t m0 x1 he y7a ff1 fs7 fc0 sc0 ls0 ws0">and interact with the people and the environment to accomplish those speciﬁc</div><div class="t m0 x1 he y7b ff1 fs7 fc0 sc0 lsc ws0">tasks.</div><div class="t m0 x1b he y7c ff1 fs7 fc0 sc0 ls0 ws0">Compared to GRACE, the standard cell phone is not too bright. Although</div><div class="t m0 x1 he y7d ff1 fs7 fc0 sc0 ls0 ws0">the common cell phone has a camera, it lacks GRACE’s vision algorithms so it</div><div class="t m0 x1 he y7e ff1 fs7 fc0 sc0 ls0 ws0">does not know what it is seeing. It can send a video clip, but it has no perception</div><div class="t m0 x1 he y7f ff1 fs7 fc0 sc0 ls0 ws0">of the visual scene in the clip. If it had GRACE-like vision algorithms, it could</div><div class="t m0 x1 he y80 ff1 fs7 fc0 sc0 ls0 ws0">perceive the visual scene. It could tell if it were at home, in the car, at work,</div><div class="t m0 x1 he y81 ff1 fs7 fc0 sc0 ls0 ws0">shopping, or driving up the driveway on the way home. If GRACE-like vision</div><div class="t m0 x1 he y82 ff1 fs7 fc0 sc0 ls0 ws0">algorithms show it that you are entering your driveway in your car, a Cognitive</div><div class="t m0 x1 he y83 ff1 fs7 fc0 sc0 lsd ws0">SDR could learn to open the garage door for you wirelessly. Thus, you would</div><div class="t m0 x1 he y84 ff1 fs7 fc0 sc0 lsd ws0">not need to ﬁsh for the garage door opener, yet another wireless gadget. In fact,</div><div class="t m0 x1 he y85 ff1 fs7 fc0 sc0 ls0 ws0">you do not need a garage door opener anymore, once CRs enter the market. To</div><div class="t m0 x1 he y86 ff1 fs7 fc0 sc0 ls0 ws0">open the car door, you will not need a key fob either. As you approach your car,</div><div class="t m0 x1 he y87 ff1 fs7 fc0 sc0 ls0 ws0">your personal CR perceives the common scene and, as trained, synthesizes the</div><div class="t m0 x1 he y88 ff1 fs7 fc0 sc0 ls0 ws0">fob RF transmission and opens the car door for you.</div><div class="t m0 x1b he y89 ff1 fs7 fc0 sc0 ls0 ws0">Your CR perceives visual scenes continuously searching visual - RF correla-</div><div class="t m0 x1 he y8a ff1 fs7 fc0 sc0 ls0 ws0">tions, cues to your needs for wireless services. A CR radio learns to open your</div><div class="t m0 x1 he y8b ff1 fs7 fc0 sc0 lse ws0">garage door when you arrive home from your use of the garage door opener.</div><div class="t m0 x1 he y8c ff1 fs7 fc0 sc0 lsd ws0">When ﬁrst you open the garage door with the wireless garage-door opener, your</div></div><div class="pi" data-data='{"ctm":[1.611850,0.000000,0.000000,1.611850,0.000000,0.000000]}'></div></div>

<div id="pf4" class="pf w0 h0" data-page-no="4"><div class="pc pc4 w0 h0"><img class="bi x0 y0 w1 h1" alt="" src="https://static.pudn.com/prod/directory_preview_static/62546b02b744eb386faa74c2/bg4.jpg"><div class="t m0 x7 h5 y28 ff1 fs1 fc0 sc0 ls2 ws0">246 Cooperation in Wireless Networks</div><div class="t m0 x7 he y60 ff1 fs7 fc0 sc0 ls0 ws0">CR correlates the visual and RF scenes: owner’s hand on device, then RF sig-</div><div class="t m0 x7 he y61 ff1 fs7 fc0 sc0 lsd ws0">nal in the ISM band, and then the garage door opens. The next time, your CR</div><div class="t m0 x7 he y8d ff1 fs7 fc0 sc0 ls0 ws0">veriﬁes through reinforcement learning that your hand on the button, the RF</div><div class="t m0 x7 he y8e ff1 fs7 fc0 sc0 lsd ws0">signal, and the opening of the garage door form a sequential script, a use-case.</div><div class="t m0 x7 he y8f ff1 fs7 fc0 sc0 lsd ws0">The third time, your cognitive radio detects the approach to the garage door and</div><div class="t m0 x7 he y90 ff1 fs7 fc0 sc0 ls0 ws0">offers to complete the RF use case for you, saying, “I see we are approaching</div><div class="t m0 x7 he y91 ff1 fs7 fc0 sc0 lsd ws0">the garage. Would you like me to open the door for us?” Thereafter, it will open</div><div class="t m0 x7 he y92 ff1 fs7 fc0 sc0 lsd ws0">the garage door when you drive up the driveway unless you tell it not to. It has</div><div class="t m0 x7 he y93 ff1 fs7 fc0 sc0 lsd ws0">transformed one of your patterns of RF usage, opening the garage door; into a</div><div class="t m0 x7 he y94 ff1 fs7 fc0 sc0 ls0 ws0">cognitive (self-user perceptive) service, ofﬂoading one of your daily tasks.</div><div class="t m0 x9 he y95 ff1 fs7 fc0 sc0 lsd ws0">Since the CR has learned to open the garage door, you may un-clutter your</div><div class="t m0 x7 he y96 ff1 fs7 fc0 sc0 ls0 ws0">car by just one widget, that door opener. Since your CR learned to open the</div><div class="t m0 x7 he y97 ff1 fs7 fc0 sc0 lsd ws0">garage door by observing your use of the radio via AML, you did not pay</div><div class="t m0 x7 he y98 ff1 fs7 fc0 sc0 ls0 ws0">the cell phone company, and you did not endure pop-up advertising to get this</div><div class="t m0 x7 he y99 ff1 fs7 fc0 sc0 ls0 ws0">personalized wireless service. As you enter the house with arms full of packages,</div><div class="t m0 x7 he y9a ff1 fs7 fc0 sc0 lsd ws0">your CR closes the garage door and locks it for you, having learned that from</div><div class="t m0 x7 he y9b ff1 fs7 fc0 sc0 ls0 ws0">you as well. For the CR vision system to see what you see, today’s Bluetooth</div><div class="t m0 x7 he y9c ff1 fs7 fc0 sc0 ls0 ws0">earpieces evolve to CR Bluetooth glasses, complete with GRACE-like vision.</div><div class="t m0 x9 he y9d ff1 fs7 fc0 sc0 ls0 ws0">CRs do not attempt everything. They learn about your radio use patterns</div><div class="t m0 x7 he y9e ff1 fs7 fc0 sc0 ls0 ws0">because they know a LOT about radio and about generic users and legitimate</div><div class="t m0 x7 he y9f ff1 fs7 fc0 sc0 ls0 ws0">uses of radio. CRs have the a-priori knowledge needed to detect opportunities to</div><div class="t m0 x7 he ya0 ff1 fs7 fc0 sc0 lsb ws0">assist you with your use of the radio spectrum accurately, delivering that assis-</div><div class="t m0 x7 he ya1 ff1 fs7 fc0 sc0 ls0 ws0">tance with minimum intrusion. TellMe</div><div class="t m0 x5 h10 ya2 ff1 fs8 fc0 sc0 ls0 ws0">R</div><div class="t m0 x1e h11 ya3 ff8 fs7 fc0 sc0 ls0 ws0">is not a generic speech understanding</div><div class="t m0 x7 he ya4 ff1 fs7 fc0 sc0 ls0 ws0">system and CR is not a generic AI service in a radio. Products realizing the</div><div class="t m0 x7 he ya5 ff1 fs7 fc0 sc0 ls0 ws0">visual perception of this vignette are realizable on laptop computers today. Re-</div><div class="t m0 x7 he ya6 ff1 fs7 fc0 sc0 ls0 ws0">inforcement learning (RL) and Case-based Reasoning (CBR) are mature AML</div><div class="t m0 x7 he ya7 ff1 fs7 fc0 sc0 ls0 ws0">technologies with radio network applications now being demonstrated in acad-</div><div class="t m0 x7 he ya8 ff1 fs7 fc0 sc0 ls0 ws0">emic and industrial research settings as technology pathﬁnders for CR [Joseph</div><div class="t m0 x7 he ya9 ff1 fs7 fc0 sc0 ls0 ws0">Mitola III, 2000a] and CWN [Petri Mahonen, 2004]. Two or three Moore’s law</div><div class="t m0 x7 he yaa ff1 fs7 fc0 sc0 ls0 ws0">cycles or three to ﬁve years from now, these vision and learning algorithms</div><div class="t m0 x7 he yab ff1 fs7 fc0 sc0 ls0 ws0">will ﬁt in your cell phone. In the interim, CWNs will begin to offer such ser-</div><div class="t m0 x7 he yac ff1 fs7 fc0 sc0 ls0 ws0">vices, offering consumers new tradeoffs between privacy and ultra-personalized</div><div class="t m0 x7 he yad ff1 fs7 fc0 sc0 lsd ws0">convenience.</div><div class="t m0 x7 hd yae ff2 fs6 fc0 sc0 ls9 ws0">Cognitive Radios Hear What You Hear, Augmenting</div><div class="t m0 x7 hd yaf ff2 fs6 fc0 sc0 ls9 ws0">Personal Skills</div><div class="t m0 x9 he yb0 ff1 fs7 fc0 sc0 ls0 ws0">Compared to GRACE, the cell phone on your waist is deaf. Although your cell</div><div class="t m0 x7 he yb1 ff1 fs7 fc0 sc0 ls0 ws0">phone has a microphone, it lacks GRACE’s speech understanding technology,</div><div class="t m0 x7 he yb2 ff1 fs7 fc0 sc0 ls0 ws0">so it does not perceive what it hears. It can let you talk to your daughter, but it</div><div class="t m0 x7 he yb3 ff1 fs7 fc0 sc0 ls0 ws0">has no perception of your daughter, nor of the content of your conversation. If it</div><div class="t m0 x7 he yb4 ff1 fs7 fc0 sc0 ls0 ws0">had GRACE’s speech understanding technology, it could perceive your speech</div></div><div class="pi" data-data='{"ctm":[1.611850,0.000000,0.000000,1.611850,0.000000,0.000000]}'></div></div>

<div id="pf5" class="pf w0 h0" data-page-no="5"><div class="pc pc5 w0 h0"><img class="bi x0 y0 w1 h1" alt="" src="https://static.pudn.com/prod/directory_preview_static/62546b02b744eb386faa74c2/bg5.jpg"><div class="t m0 x1 h5 y28 ff3 fs5 fc0 sc0 ls8 ws0">Cognitive Radio Architecture 247</div><div class="t m0 x1 he y60 ff1 fs7 fc0 sc0 ls0 ws0">dialog. It could detect that you and your daughter are talking about common</div><div class="t m0 x1 he yb5 ff1 fs7 fc0 sc0 ls0 ws0">subjects like homework, or your favorite song. With CR, GRACE-like speech</div><div class="t m0 x1 he yb6 ff1 fs7 fc0 sc0 ls0 ws0">algorithms would detect your daughter saying that your favorite song is now</div><div class="t m0 x1 he y8e ff1 fs7 fc0 sc0 ls0 ws0">playing on WDUV. As an SDR, not just a cell phone, your CR tunes to FM 105.5</div><div class="t m0 x1 he yb7 ff1 fs7 fc0 sc0 ls0 ws0">so that you can hear “The Rose.” With your CR, you no longer need a transistor</div><div class="t m0 x1 he yb8 ff1 fs7 fc0 sc0 ls0 ws0">radio. Your CR eliminates from your pocket, purse or backpack yet another RF</div><div class="t m0 x1 he yb9 ff1 fs7 fc0 sc0 lse ws0">gadget. In fact, you may not need iPOD</div><div class="t m0 x1f h10 y69 ff1 fs8 fc0 sc0 ls0 ws0">R</div><div class="t m0 x20 h11 y6a ff8 fs7 fc0 sc0 ls0 ws0">, GameBoy</div><div class="t m0 x21 h10 y69 ff1 fs8 fc0 sc0 ls0 ws0">R</div><div class="t m0 x22 h11 y6a ff8 fs7 fc0 sc0 ls0 ws0"> and similar products</div><div class="t m0 x1 he yba ff1 fs7 fc0 sc0 ls0 ws0">as high-end CR’s enter the market. Your CR will learn your radio listening and</div><div class="t m0 x1 he ybb ff1 fs7 fc0 sc0 ls0 ws0">information use patterns, accessing the songs, downloading games, snipping</div><div class="t m0 x1 he ybc ff1 fs7 fc0 sc0 ls0 ws0">broadcast news, sports, stock quotes as you like as the CR re-programs its</div><div class="t m0 x1 he ybd ff1 fs7 fc0 sc0 ls0 ws0">internal SDR to better serve your needs and preferences. Combining vision and</div><div class="t m0 x1 he ybe ff1 fs7 fc0 sc0 ls0 ws0">speech perception, as you approach your car your CR perceives this common</div><div class="t m0 x1 he ybf ff1 fs7 fc0 sc0 ls0 ws0">scene and, as you had the morning before, tunes your car radio to WTOP to your</div><div class="t m0 x1 he yc0 ff1 fs7 fc0 sc0 ls0 ws0">favorite “Trafﬁc and weather together on the eights.” With GRACE’s speech</div><div class="t m0 x1 he yc1 ff1 fs7 fc0 sc0 ls0 ws0">understanding algorithms, your CR recognizes such regularly repeated catch</div><div class="t m0 x1 he yc2 ff1 fs7 fc0 sc0 ls0 ws0">phrases, turning up the volume for the trafﬁc report and then turning it down</div><div class="t m0 x1 he yc3 ff1 fs7 fc0 sc0 ls0 ws0">or off after the weather report, avoiding annoying commercials and selecting</div><div class="t m0 x1 he yc4 ff1 fs7 fc0 sc0 ls0 ws0">relevant ones. If you actually need a tax deduction, it will record those radio</div><div class="t m0 x1 he yc5 ff1 fs7 fc0 sc0 ls0 ws0">commercials for your listening pleasure at tax time when you need them.</div><div class="t m0 x1b he yc6 ff1 fs7 fc0 sc0 ls0 ws0">For AML, CRs need to save speech, RF, and visual cues, all of which may</div><div class="t m0 x1 he yc7 ff1 fs7 fc0 sc0 ls0 ws0">be recalled by the user, expanding the user’s ability to remember details of con-</div><div class="t m0 x1 he yc8 ff1 fs7 fc0 sc0 ls0 ws0">versations and snapshots of scenes, augmenting the skills of the <Owner/></div><div class="t m0 x23 h12 yc9 ffa fs3 fc0 sc0 ls0 ws0">1</div><div class="t m0 x24 he yca ff1 fs7 fc0 sc0 ls0 ws0">.</div><div class="t m0 x1 he ycb ff1 fs7 fc0 sc0 ls0 ws0">Because of the brittleness of speech and vision technologies, CRs try to “re-</div><div class="t m0 x1 he ycc ff1 fs7 fc0 sc0 ls0 ws0">member everything” like a continuously running camcorder. Since CRs detect</div><div class="t m0 x1 he ycd ff1 fs7 fc0 sc0 ls0 ws0">content such as speakers’ names, and keywords like “radio” and “song,” they</div><div class="t m0 x1 he yce ff1 fs7 fc0 sc0 ls0 ws0">can retrieve some content asked for by the user, expanding the user’s memory</div><div class="t m0 x1 he ycf ff1 fs7 fc0 sc0 ls0 ws0">in a sense. CRs thus could enhance the personal skills of their users such as</div><div class="t m0 x1 he yd0 ff1 fs7 fc0 sc0 ls0 ws0">memory for detail.</div><div class="t m0 x1b he yd1 ff1 fs7 fc0 sc0 ls0 ws0">High performance dialog and audio-video retrieval technologies are cutting-</div><div class="t m0 x1 he yd2 ff1 fs7 fc0 sc0 ls0 ws0">edge but not out of reach for suitably narrow domains like TellMe</div><div class="t m0 x25 h10 yd3 ff1 fs8 fc0 sc0 ls0 ws0">R</div><div class="t m0 x26 h11 yd4 ff8 fs7 fc0 sc0 ls0 ws0"> and cus-</div><div class="t m0 x1 he yd5 ff1 fs7 fc0 sc0 ls0 ws0">tomization of wireless services. Casual dialog typically contains anaphora and</div><div class="t m0 x1 he yd6 ff1 fs7 fc0 sc0 ls0 ws0">ellipsis, using words like “this” and “that” to refer to anonymous events like</div><div class="t m0 x1 he yd7 ff1 fs7 fc0 sc0 ls0 ws0">playing a favorite song. Although innovative, speech research systems already</div><div class="t m0 x1 he yd8 ff1 fs7 fc0 sc0 ls0 ws0">achieve similar dialogs in limited domains [Victor Zue, 2005]. When the user</div><div class="t m0 x1 he yd9 ff1 fs7 fc0 sc0 ls0 ws0">says, “How did you do that?” the domain of discourse is limited to the <Self/></div><div class="t m0 x1 he yda ff1 fs7 fc0 sc0 ls0 ws0">and its contemporaneous actions. Since CR can do only one or two things at</div><div class="t m0 x1 he ydb ff1 fs7 fc0 sc0 ls0 ws0">once, the question, “How did you do that?” has only one primary semantic ref-</div><div class="t m0 x1 he ydc ff1 fs7 fc0 sc0 ls0 ws0">erent, playing the song. Reasoning using analogy, also cutting edge, is no longer</div><div class="t m0 x1 he ydd ff1 fs7 fc0 sc0 ls0 ws0">beyond the pale for tightly limited domains like CR and thus is envisioned in</div><div class="t m0 x1 he yde ff1 fs7 fc0 sc0 lse ws0">the CRA.</div></div><div class="pi" data-data='{"ctm":[1.611850,0.000000,0.000000,1.611850,0.000000,0.000000]}'></div></div>