Telecommunications Networking I Topic 1 Overview of Telecommunications Networking I-II Dr. Stewart D. Personick Drexel University sdp@ece.drexel.edu 1 Copyright 2001, S.D. Personick. All rights Telecommunications Networking • Sending and receiving messages (couriers, smoke signals, flashes of light, telegraph, teletype, facsimile, voice mail, text E-mail, multimedia E-mail) • Real-time conversations and collaborative work (face-to-face, wireline telephone, 2-way radio-telephone, video teleconferencing, multimedia teleconferencing) • Accessing stored or real-time information (physical file cabinets and folders, FTP, WWW, remote sensing, imagery) 2 • Networked Copyright information systems (mission2001, S.D. Personick. All rights Telecommunications Networking • How to communicate in efficient, predictable, reliable, and useful ways? -Representing data or information as a “signal” -Defining the quality (or fidelity) of reception/communication -Signals in the presence of noise, distortion and interference -Point-to-point communication systems 3 -Switched networks (e.g., the Internet, LANs, Copyright 2001, S.D. Personick. All rights • Overview of Telecommunications Networking I Overview of telecommunications Networking I-II • Data, information (voice, audio, images, video), and signals that represent data and information • Signals in noise, measures of quality of communication • Wire-pair and coaxial cable, optical fiber, and wireless transmission systems (link layer) -radio frequency wireless links -lasercom (free-space optical) wireless 4 Copyright 2001, S.D. Personick. All rights links Overview of Telecommunications Networking II • Wireless systems and networks: peer-to-peer communication using a shared “ether”, broadcast systems, cellular/PCS, 2-way satellite systems • Local area networks (packet switching) • The Internet • Battlespace systems • Next generation systems, networks, issues and applications 5 Copyright 2001, S.D. Personick. All rights Telecommunications Networking I Topic 2 Data and Information, and Signals that Represent Data and Information Dr. Stewart D. Personick Drexel University 6 Copyright 2001, S.D. Personick. All rights Data, Information and Signals • Sound: speech, audio, represented, modeled and simulated • Text and Images: character mapped, scanned, bit mapped, transform representations • Video: real and animated, frame-byframe, interframe compression coded 7 Copyright 2001, S.D. Personick. All rights Capturing Sound 8 Copyright 2001, S.D. Personick. All rights Capturing Sound • Sound takes the physical form of an acoustic wave... i.e., variations in pressure vs time and space... that travels through a compressible physical medium such as air (~1090 feet/second… ~332 meters/second) • A microphone (transducer) converts locally received pressure variations into a varying voltage/current that 9 Copyright 2001, S.D. Personick. All rights Capturing Sound (cont’d) • The varying voltage waveform that represents the captured sound is communicated to another location using one of many possible communication system technologies • The received varying voltage waveform is not an exact replica of the transmitted voltage waveform 10 Copyright 2001, S.D. Personick. All rights Capturing Sound (cont’d) • The received varying voltage waveform is used to “drive” a speaker (transducer) which produces a new acoustic wave (sound) that is perceived as an approximation of the original sound • Does the reproduced acoustic wave “sound” like the original acoustic wave? 11 Copyright 2001, S.D. Personick. All rights Capturing Sound (cont’d) • The received varying voltage waveform is used to “drive” a speaker (transducer) which produces a new acoustic wave (sound) that is perceived as an approximation of the original sound • Does the reproduced acoustic wave “sound” like the original acoustic wave? The answer depends upon the application 12 Copyright 2001, S.D. Personick. All rights Representing Speech 13 Copyright 2001, S.D. Personick. All rights Representing Speech (cont’d) • Speech is one of the most important analog signals • Representation qualities include: -Intelligibility: Can I understand what you are saying? Can I build a machine that responds properly to what you are saying? -Naturalness: Does it sound like face-toface communication? Can I identify the 14 speaker?Copyright 2001, S.D. Personick. All rights Representing Speech (cont’d) • Traditional telephone quality speech: 3 kHz high frequency cutoff, small amounts of noise and echo • AM radio quality speech: 5 kHz high frequency cutoff, varying amounts of noise and interference • FM radio, TV, other high quality speech: 10 kHz+ high frequency cutoff, barely 15 perceptible noise Copyright 2001, S.D. Personick. All rights Representing Speech (cont’d) • Traditional telephone quality speech: 3 kHz high frequency cutoff, small amounts of noise and echo [ ~32 – 64 kbps ] • AM radio quality speech: 5 kHz high frequency cutoff, varying amounts of noise and interference [~64 – 128 kbps] • FM radio, TV, other high quality speech: 10 kHz+ high frequency cutoff, barely 16 perceptible noise Copyright 2001, S.D. Personick. All rights Representing Speech (cont’d) • Compressed speech -Uses digital signal processing to remove redundancies in the original speech signal. -This typically impacts on the naturalness and comfort associated with the speech signal produced at the receiving end of a link, but (hopefully) still provides intelligibility 17 Copyright 2001, S.D. Personick. All rights Representing Speech (cont’d) • Compressed speech [~8 – 16 kbps] -Uses digital signal processing to remove redundancies in the original speech signal. -This typically impacts on the naturalness and comfort associated with the speech signal produced at the receiving end of a link, but (hopefully) still provides intelligibility 18 Copyright 2001, S.D. Personick. All rights Representing Speech (cont’d) • Modeled and simulated speech -Uses a model of the vocal tract to generate sounds that are perceived as intelligible speech -Real speech is captured by estimating the vocal tract parameters (that vary relatively slowly in time v. the sound speech waveform itself). These parameters are stored and/or 19 transmitted, and used to recreate the Copyright 2001, S.D. Personick. All rights Representing Speech (cont’d) • Modeled and simulated speech [~1.2 – 4.8 kbps] -Uses a model of the vocal tract to generate sounds that are perceived as intelligible speech -Real speech is captured by estimating the vocal tract parameters (that vary relatively slowly in time v. the sound speech waveform itself). These 20 parametersCopyright are2001, stored and/or S.D. Personick. All rights Representing Audio • Audio signals, like music typically demand a high accuracy of representation to meet users’ expectations >10 kHz high frequency cutoff <100 Hz low frequency cutoff low noise and distortion • A typical audio system specification includes a 20-20,000 Hz “frequency response” Copyright 2001, S.D. Personick. All rights 21 Representing Audio • Audio signals, like music typically demand a high accuracy of representation to meet users’ expectations >10 kHz high frequency cutoff <100 Hz low frequency cutoff low noise and distortion • A typical audio system specification includes a 20-20,000 Hz “frequency response” [128 kbps – 1 Mbps] 22 Copyright 2001, S.D. Personick. All rights Capturing Analog Images • Use a camera or scanner (transducer) to produce a signal or a set of data which represents the image • Communicate this signal or data to a receiving location • Use the received signal or data [which is not necessarily identical to the transmitted signal or data] to reconstruct 23 a new image Copyright 2001, S.D. Personick. All rights Representing Images 24 Copyright 2001, S.D. Personick. All rights Representing Images • Character-mapped Images -The image consists of a number of “characters” or objects selected from a data base -To capture the image, one must obtain or derive its description in the form of: data that represents each character or object used; data representing its location on the image; data describing colors used, fonts, object sizes, object and character overlaps, etc. -The set of data is communicated to the 25 receiving location and used to recreate the Copyright 2001, S.D. Personick. All rights Character Mapped Images Figure 1 Figure 1 26 Copyright 2001, S.D. Personick. All rights Representing Images • Scanned images -Scan the image …e.g., left-to-right, and topto-bottom -Represent the scanned brightness and color (e.g., red, green, and blue color brightness) by a set of signals which change in time as the scanning point moves -Communicate these signals to the receiving location; and use them to “paint” a new image with a complementary scanning process 27 Copyright 2001, S.D. Personick. All rights Scanned Images Scan 28 Copyright 2001, S.D. Personick. All rights Representing Images • Bit-mapped Images -Divide the image into an n x m array of “pixels” (e.g., 800 x 600 = 480,000 pixels) -Represent the brightness and color of each pixel (e.g., red, green, and blue color intensities) by a set of numbers -Communicate these numbers to the receiving end, and use them to recreate the image 29 Copyright 2001, S.D. Personick. All rights Bit-mapped Images 30 Copyright 2001, S.D. Personick. All rights Bit Mapped Image Example An image contains 800 x 600 = 480,000 pixels 3 bytes of information are required to represent the intensity and color of each pixel To store this image you would require 480,000 x 3 bytes of memory = 1,440,000 Bytes To transmit this image in 1 second, you must transmit at a data rate of 480,000 (pixels) x 3 (bytes per pixel) x 8 (bits per byte) bits per second = 11,520,000 bits per second Copyright 2001, S.D. Personick. All rights 31 Representing Images • Transforms -Example: Hadamard transform (for a single color) What is the average brightness across the entire image? What is the difference in the brightness of the upper left quadrant vs the upper right quadrant? Upper left vs lower left? Upper right 32 Copyright 2001, S.D. Personick. All rights vs lower right? Transform Image Representation 33 Copyright 2001, S.D. Personick. All rights Transforms • Creative use of transform coding can reduce the amount of information required to represent an image. Example -if the intensity and color of an image isn’t changing across a selected portion of the image, then a single set of intensity and color data (e.g., 3-12 bytes) plus some location information can represent that entire portion of the image • Typical compression achieved: 10 – 20 : 1 34 Copyright 2001, S.D. Personick. All rights [less for “busy images” where details must be Video 35 Copyright 2001, S.D. Personick. All rights Video (continued) • A video is a sequence of images (called “frames”), typically presented to a viewer at 24-80 frames per second • If the images within the video are captured by scanning, then each image may be scanned twice, with “interlaced” scans. Thus each image may be represented by two interlaced “fields”. (e.g., NTSC video uses 60 interlaced fields per second = 30 frames per second). 36 Copyright 2001, S.D. Personick. All rights Interlacing Field #1 Field #2 37 Copyright 2001, S.D. Personick. All rights Video (continued) • Given a fixed number of scan lines per second, the use of interlacing allows one to increase the field rate-- to reduce the perceptual artifact called “flicker”-- while maintaining a high enough number of scan lines per frame. • For some applications (e.g., computer displays of multimedia information) the use of “progressive scanning” (no interlacing) is preferred. • Experts disagree of the relative merits of 38 2001, S.D. Personick. All rights progressiveCopyright scanning vs interlaced scanning Video (continued) • While a traditional video signal is generated by a video camera that scans the (moving) images formed on its focal plane, many modern “videos” may be created using computer generated video (i.e., animations, like the original Walt Disney animations), where each image in the video sequence is intrinsically a bit mapped image or some standard compressed image file 39 Copyright 2001, S.D. Personick. All rights NTSC Video Scan lines per frame = 525 (262.5 x 2, interlaced) Frame rate = 30 frames per second Field rate = 60 fields per second Scan rate = 15,750 “lines” per second Bandwidth of scanned signal (B&W information) = 6 MHz Data rate required (no compression) ~ 96 Mbps (B&W) – 288 Mbps (3-component color) Copyright 2001, S.D. Personick. All rights 40 HDTV Video 1100 lines per frame ( v. 525 lines per frame for NTSC) 16 : 9 Aspect ratio (v. 4 : 3 for NTSC) Frame rate = 30 frames per second Bandwidth of scanned signal ~ 30 MHz per color (3 colors) Data rate required (no compression) ~ 1.44 Gbps 41 Copyright 2001, S.D. Personick. All rights Video Compression Coding • Remove redundant information within each frame, as in image compression coding • Remove redundancy that exists from frame-to-frame by -communicating only the differences that exist from one frame (or several prior frames) to the next -employing motion Copyright 2001, S.D. Personick. All rights 42 Video Compression Coding • Remove redundant information within each frame, as in image compression coding • Remove redundancy that exists from frameto-frame by -communicating only the differences that exist from one frame (or several prior frames) to the next -employing motion prediction/compensation for moving objects • AchievableCopyright compression: ~100:1 2001, S.D. Personick. All rights 43 Telecommunications Networking I Topic 3 Quantifying the Performance of Communication Systems Carrying Analog Information Dr. Stewart D. Personick Drexel University 44 Copyright 2001, S.D. Personick. All rights Signals in Noise The basic model: data or information signal + data or information signal + noise noise 45 Copyright 2001, S.D. Personick. All rights Analog Signals in Noise: Example noise Engine Temperature Sensor + s= ca (volts) where a= temperature (C) r=s+n =ca+n c= .01 volt/degree-C n(av) =0, var(n) = .0001 volt**2 46 Copyright 2001, S.D. Personick. All rights The distribution of the heights of Vulcans children 12” 24 ” adult males 36” 48 ” adult females 60” 72” 47 Copyright 2001, S.D. Personick. All rights Analog Signals in Noise Example (continued) a= a number representing information to be communicated. apriori, a is a Gaussian random variable with variance A, and zero average value s = a signal in the form of a voltage proportional to a …. = ca (volts), where c is a known constant r = a received signal with additive noise = s + n (volts) n = a Gaussian random variable with variance N (volts**2) How can we estimate “a”, if we receive “r”? Copyright 2001, S.D. Personick. All rights 48 Signals in Noise • What is the criterion for determining whether we’ve done a good job in estimating “a” from the received signal “r”? • It would have something to do with the difference between our estimated value for “a” and the true value for “a” • Example: minimize E{(a-a)**2}, where a is the estimated value of “a”, given r 49 Copyright 2001, S.D. Personick. All rights Signals in Noise • We will give a proof, on the blackboard, that a, the estimate of “a” that minimizes the average value of (a-a)**2 is a(r) = E (a|r) = the “expected value” of “a”, given “r” • The above is true for any probability distributions of “a” and “n”; and for the specific cases given, a(r) = r/c [Ac**2/(Ac**2 +N)] Copyright 2001, S.D. Personick. All rights 50 Signals in Noise Harder example: a = a Gaussian random variable with variance A, representing information s(t) = a signal of duration T (seconds) where s(t) = a c(t), and c(t) is a known waveform r(t) = a received signal = s(t) + n(t), where n(t) is a “random process” having a set of known statistical characteristics How do we estimate a, given r(t)? 51 Copyright 2001, S.D. Personick. All rights Signals in Noise • What is the criterion for evaluating how good an estimate of “a” we have derived from r(t)? • How do we describe the noise n(t) in a way that is useful in determining how to estimate “a” from r(t)? 52 Copyright 2001, S.D. Personick. All rights Signals in Noise • Suppose n(t) = nc(t) + x(t) where: n is a Gaussian random variable of variance N; and where, in some loosely defined sense: • x(t) is a random process that is statistically independent of the random variable “n”, and where …. (continued on next slide) 53 Copyright 2001, S.D. Personick. All rights Signals in Noise • ...x(t) is also “orthogonal” to the known waveform c(t), then we can construct a “hand waiving” argument that suggests that we can ignore x(t), and concentrate on the noise nc(t), as we attempt to estimate the underlying information variable “a”. • To make this argument more precisely requires a deep understanding of the 54 theory of random processes Copyright 2001, S.D. Personick. All rights Signals in Noise • We will show (using the blackboard) that we can convert this problem into the earlier problem of estimating an information parameter “a” from a received signal of the form r= ca + n. • While doing so, we will introduce the concept of a “matched filter” 55 Copyright 2001, S.D. Personick. All rights R(t) = a c(t) + n(t) [0,T] But : N(t) = n c(t) + x(t) R(t) = (a + n) c(t) + x(t) Integral of { r(t) c(t)} = Integral { (a + n) c(t) c(t) + x(t) c(t) } 56 Copyright 2001, S.D. Personick. All rights Gaussian Random Processes • If we look at (“sample”) the random process n(t) at times: t1,t2,t3,…,tj, then we get a set of random variables: n(t1), n(t2), n(t3), …,n(tj). • If the set of random variables {n(tj)} has a joint probability density that is Gaussian, then n(t) is called a Gaussian random process 57 Copyright 2001, S.D. Personick. All rights Gaussian Random Processes (cont’d) • Any linear combination of samples of a Gaussian random process is a Gaussian random variable • Extending the above, the integral of the product n(t)c(t) over a time interval T is also a Gaussian random variable if, n(t) is a Gaussian random process, and c(t) is a known function 58 Copyright 2001, S.D. Personick. All rights Gaussian Random Processes (cont’d) • Let n(t) be a random process (not necessarily Gaussian) • Define “n” as follows: n= the integral over T of n(t)c(t)/W, where W= the integral over T of c(t)c(t) • Then, we can write n(t) as follows: n(t)= nc(t) + “the rest of n(t)” 59 Copyright 2001, S.D. Personick. All rights Gaussian Random Processes (cont’d) • If n(t) is a “white, Gaussian random process”, then: -n is a Gaussian random variable, and - “rest of n(t)” is statistically independent of “n”…I.e., “the rest of n(t)” contains no information that can help of estimate either “n” or “a” 60 Copyright 2001, S.D. Personick. All rights Gaussian Random Processes (cont’d) • Furthermore, we can build a correlator that works as follows. It takes the received waveform, r(t), multiplies it by the known waveform c(t), integrates over the time interval T, and finally divides by W z= {the integral over T of r(t)c(t)}/W 61 Copyright 2001, S.D. Personick. All rights Gaussian Random Processes (cont’d) • Going back to the definitions and equations above, we find that z= a + n, where “a” is the original information variable we wish to estimate, and n is a Gaussian random variable • Thus by introducing the correlator, we convert the new problem to (continued)62 Copyright 2001, S.D. Personick. All rights Gaussian Random Processes (cont’d) • …the old problem of estimating a scalar (“a”) from another scalar (“r”); where r= a+n, and where n is a Gaussian random variable • The correlation operation is also known as “matched filtering”, because it can be accomplished by passing r(t) through a filter whose impulse response is c(-t). 63 Copyright 2001, S.D. Personick. All rights Capturing analog waveforms 64 Copyright 2001, S.D. Personick. All rights Example Information Waveform Time 65 Copyright 2001, S.D. Personick. All rights Example Pulse stream Time 66 Copyright 2001, S.D. Personick. All rights Example Sampling 67 Copyright 2001, S.D. Personick. All rights Example Samples 68 Copyright 2001, S.D. Personick. All rights Example Pulse Amplitude Modulation (PAM) 69 Copyright 2001, S.D. Personick. All rights Example PAM Stream 70 Copyright 2001, S.D. Personick. All rights Example The PAM stream is a representation of the information signal 71 Copyright 2001, S.D. Personick. All rights Example s(t) = transmitted signal 72 Copyright 2001, S.D. Personick. All rights Example r(t) = s(t) + n(t) = received signal 73 Copyright 2001, S.D. Personick. All rights Matched Filtering y (t) a x(t) h(t)= x(-t) Matched Filter y(t)= integral of [h(t-u) a x(u)du] = integral of [ x(u-t) a x(u)du] y(0)= integral of [a (x(u)x(u)du] = a E 74 Copyright 2001, S.D. Personick. All rights Matched Filtering y (t) a x(t) + n(t) h(t)= x(-t) Matched Filter y(t)= integral of [h(t-u) {a x(u) + n(u)}du] = integral of [ x(u-t) {a x(u) + n(u)}du] y(0)= integral of [ (x(u){a x(u) + n(u)}du] = a E + n 75 Copyright 2001, S.D. Personick. All rights Example • If each of the random variables “a” is a Gaussian random variable with variance A • and, if n(t) is white Gaussian noise with “spectral density” N • Then the optimal estimate of each of the variables “a” is given by “sampling” the output of the matched filter and 76 multiplyingCopyright it by: (1/E) [AE/(N+AE)] 2001, S.D. Personick. All rights Example • If each of the random variables “a” is a Gaussian random variable with variance A • and, if n(t) is “white” Gaussian noise with “spectral density” N • Then the associated mean squared error of each estimate will be A [N/(N+AE)] 77 Copyright 2001, S.D. Personick. All rights Telecommunications Networking I Topic 4 Point-to-Point Communication over Metallic Cables Dr. Stewart D. Personick Drexel University 78 Copyright 2001, S.D. Personick. All rights Metallic Cables Insulator Insulated Copper Wires Shield Coaxial Cable Center Conductor 79 Copyright 2001, S.D. Personick. All rights Metallic Cables • Copper wire (“wire pair”) cable is a very important medium in the commercial sector, because -essentially all homes and small businesses in the United States, and in many other parts of the world, currently access the worldwide telecommunications infrastructure using a pair of wires -wire pairs are very commonly used for local area networks in offices and 80 Copyright 2001, S.D. Personick. All rights Metallic Cables (cont’d) • Coaxial cable is a very important medium in the commercial sector, because -more than 80% of residences in the U.S., and a large fraction of residences in many other parts of the world can access the telecommunications infrastructure using coaxial cable -many local area networks utilize 81 coaxial cable Copyright 2001, S.D. Personick. All rights Metallic Cables (cont’d) • While optical fibers and wireless carry a growing share of telecommunications traffic in the commercial sector, metallic cables will continue to carry the majority of local area network and access network traffic for many years to come • The cost of replacing all of the metallic access cable in the U.S. with fiber would be around $1000/home x 100 82 million homes Copyright 2001, S.D. Personick. All rights Metallic Cable System Informatio n cable s(t) Transmitter Signal (e.g., 1 volt peak) r(t) Receiver Information 83 Copyright 2001, S.D. Personick. All rights What can we say about r(t)? • r(t) = s(t)*h(t) + n(t) + i(t), where: s(t) is the signal that enters the cable, h(t) is the impulse response of the cable, n(t) is noise associated with the finite temperature of the cable, i(t) is interference from other signals, and a*b means: the convolution of a(t) and Copyright 2001, S.D. Personick. All rights 84 What can we say about h(t)? • h(t) is the cable’s impulse response, and is equal to the Fourier transform of the cable’s frequency response: H(f) • H(f), the cable’s frequency response, approximately takes the form: 20 log H(f)= -aL[f**(1/2)], where L is the cable length (km) and a is a constant [dB/km-Hz**(1/2)] 85 Copyright 2001, S.D. Personick. All rights Typical Coaxial Cable Losses 14 12 10 RG-174 RG-58 RG-8 8 dB/100 feet 6 4 2 0 1 MHz 10 MHz 100 1 GHz MHz 86 Copyright 2001, S.D. Personick. All rights What can we say about h(t)? (cont’d) • The frequency response, H(f), “rolls off” as 10**[- square root of the frequency: f], due to the “skin effect” in a metallic cable (either wire pair cable or coaxial cable) • This roll off, by definition, attenuates higher frequencies more than lower frequencies; and causes “dispersion” of the signal: s(t) 87 Copyright 2001, S.D. Personick. All rights “Dispersion” Input pulse volts vs. time (microseconds) Output pulse millivolts vs. time (microseconds) 88 Copyright 2001, S.D. Personick. All rights What can we say about n(t)? • Noise, as we define it here, is an unwanted signal, or the sum of several unwanted signals….each of which is caused by a natural phenomenon • Examples of sources of noise are: -thermally induced, random fluctuations of the properties of materials (thermal noise) -lightning 89 Copyright 2001, S.D. Personick. All rights What can we say about n(t)? • In cables, n(t) is usually well modeled as white Gaussian noise • This additive noise is called “thermal noise”, and results from the combination of the finite temperature of the cable (e.g., 293K) and the finite loss of the cable 90 Copyright 2001, S.D. Personick. All rights What can we say about i(t)? • i(t) is a man made interference signal that adds to the desired signal at the receiver • Interference can be caused by: -signals on other pairs of wire in the same wire pair cable (called “crosstalk”) -signals generated outside of the cable that “leak” into the cable (e.g., nearby, strong radio signals) 91 Copyright 2001, S.D. Personick. All rights Our challenge: • Figure out how to do the best job we can of recovering the underlying information being communicated, given the received signal r(t) • Understand what the basic limitations of cable systems are: e.g.; how far can we transmit and still recover the underlying information with adequate fidelity? 92 Copyright 2001, S.D. Personick. All rights Example: Dispersion-Limited Operation; 6dB maximum roll off s(t) r(t) cable s(t) is a 1 volt pulse, 100 ns wide; The cable is RG8X, with a loss of 1.8 dB/100ft @ 20MHz; Suppose that the maximum allowable loss at 1/(100ns) = 10 MHz is 6dB What is the maximum allowable cable length? Copyright 2001, S.D. Personick. All rights 93 Example: Dispersion Limited Operation (cont’d) • If the cable loss at 20 MHz is 1.8 dB/100ft, then the loss at 10 MHz is 1.8 x [(10/20)**0.5)] = 1.8 [.707] ~ 1.3dB/100ft • If the maximum allowable roll off, at 10MHz, between the transmitter and the receiver is 6dB, then the maximum allowable cable length is (6/1.3) (100ft)~ 460 feet 94 Copyright 2001, S.D. Personick. All rights Example: Noise Limited Operation Cable: H(f) G(f) s(t) Noise=4kTB Receiver (includes equalizer) Equalizer: G(f) H(f): 95 Copyright 2001, S.D. Personick. All rights Noise Limited Operation (cont’d) • The “equalizer” in the receiver amplifies higher frequencies more than lower frequencies, to compensate for the “roll off” in the cable • The equalizer also amplifies the higher frequency noise at its input • kT= Boltzman’s constant x temperature (K) = approximately 4 x 10**-21 (J) @ 293K 96 Copyright 2001, S.D. Personick. All rights Example: Noise Limited Operation • Suppose the receiver has input noise, within band B, whose variance is 4kTB (watts), where B= 1/100ns • Suppose the signal-to-noise ratio (SNR) at the receiver input must be > 100:1 = 20dB; where: SNR=[{s(max)**2}/R] [10**-(aL/10)]/[4kTB] • Suppose: s(max) = 1 volt; a = 5dB/100ft; T= 293K, and R= 50 (ohms) • What is the maximum allowable cable length L? 97 Copyright 2001, S.D. Personick. All rights Example: Noise Limited Operation (cont’d) • [s(max)**2]/R= [(1)**2]/50 = 0.02 watts • 4kTB = 4 (4 x10**-21)(10**7)= 1.6x10**-13 watts • SNR= [.02/(1.6x10**-13)][10**-(aL/10)] = 1.25x10**11 [10**-(aL/10)] • If the SNR must be greater than 100 (I.e., 20dB), then [10**(-aL/10)] must be greater than 1.25x10**-9; I.e., aL<89dB 98 Copyright 2001, S.D. Personick. All rights Example: Noise Limited Operation (cont’d) • If aL must be less than 89 dB, and a equals 5 dB/100ft, then the maximum cable length is 89/5 (100 ft) ~ 1800 ft 99 Copyright 2001, S.D. Personick. All rights Line Amplifiers • We can extend the distance of transmission in a noise-limited situation, by the use of line amplifiers • The underlying concept is the equalize and amplify the signal before it has become attenuated by the cable to the point where the signal-to-noise ratio drops below an acceptable level: 100 Copyright 2001, S.D. Personick. All rights Line Amplifiers (cont’d) Transmitter Cable section #1 Line Amplifier #1 Cable section #2 Line Amplifier #2 Cable section #3 101 Copyright 2001, S.D. Personick. All rights Line Amplifiers (cont’d) • Each line amplifier equalizes and amplifies its input signal to compensate the the roll off (frequency dependent loss) of the previous section of cable • Each line amplifier also amplifies the noise arriving at its input, and adds some additional noise of its own 102 Copyright 2001, S.D. Personick. All rights G(f) Line Amplifiers (cont’d) Cable section: H(f) Gain= G(f)~ 1/[H(f)] frequency 103 Copyright 2001, S.D. Personick. All rights Line Amplifiers (cont’d) • If the gain vs. frequency of the line amplifier approximately cancels the loss vs. frequency of the cable section preceding it…I.e., if G(f)~[1/H(f)] ….. Then the net gain through the combination of the cable section and the line amplifier is 1 (0 dB) over the range of frequencies where the above relationship holds 104 Copyright 2001, S.D. Personick. All rights Line Amplifiers (cont’d) • If the noise power per unit frequency (watts per Hz) arriving at the input of the first line amplifier is defined as Nin1(f), then the noise power per unit frequency at the output of that amplifier is: Nout1 (f) = {Nin1(f) [G(f)]**2} + Ninamp(f)[G(f)**2]}; where Ninamp(f) is noise added by the 105 amplifier Copyright 2001, S.D. Personick. All rights Line Amplifiers (cont’d) • The noise power at the output of the next cable section, which is also the noise arriving at the input to the following amplifier is (approximately): Nin2(f)=Nout1(f)[H(f)]**2 + Ncable(f)= Nin1(f) + Ninamp(f) + Ncable(f); where Ncable(f) is noise added by the cable section 106 Copyright 2001, S.D. Personick. All rights Line Amplifiers (cont’d) • Furthermore, if we extend this approach to a cascade of K amplifiers and K+1 sections of cable, then the noise arriving at the input to the Kth line amplifier is: NinK(f) = Nin1(f) + [K-1] [ ninamp(f) + ncable(f) ]; • Thus in a cascade of K amplifiers, the noise arriving at the input of the Kth amplifier is approximately proportional 107 Copyright 2001, S.D. Personick. All rights Line Amplifiers (cont’d) • If cable noise from the first section of cable dominates Nin1(f), and if the receiver adds input noise comparable to that of a line amplifier….then the total noise after K sections of cable, including the noise of the receiver is: NtotalK(f) = K [ Ncable(f) + Ninamp(f)] 108 Copyright 2001, S.D. Personick. All rights Line Amplifiers (cont’d) • The result of the above is: If the signal-to-noise ratio at the input of the final line amplifier (preceding the extraction of the arriving information from the arriving signal) must be greater than some threshold, SNRmin… then the allowable loss in each section of cable decreases as more sections are added 109 Copyright 2001, S.D. Personick. All rights Line Amplifiers (cont’d) • Quantitatively…If the allowable loss between the transmitter and the receiver in a noise-limited single segment transmission system is X dB, then the allowable loss in each segment of a transmission system with K sections, and K amplifiers/equalizers (including the receiver) is X-[10logK] dB 110 Copyright 2001, S.D. Personick. All rights Line Amplifiers (continued) • In a previous example of a loss limited, single segment transmission system, the allowable cable loss (at 10 MHz) was 89dB If we try to extend the reach of the system by using 10 cable segments, then the allowable loss in each segment is 79dB; and the total loss of all the segments can be as much as 790dB (less than 10 x 89dB) 111 Copyright 2001, S.D. Personick. All rights Traditional Cable Television System Head end Super trunk Line amplifier Trunk Splitter Feeder Drop Home 112 Copyright 2001, S.D. Personick. All rights Frequency Division Multiplexing (FDM): Typical Cable System Video signals (each 6MHz wide) spaced every 6 MHz* *except for a 4MHz gap above ch4, and a 32MHz gap above ch6 Ch 2: 57 MHz +/- 3MHz Ch 75: 531MHz +/- 3 MHz 500 f (MHz) 88-120 MHz 113 Copyright 2001, S.D. Personick. All rights FDM: Typical Cable System • If there are 75 television channels sharing the same composite signal (stacked up in frequency), and if distortion considerations limit the output of the transmitter or a line amplifier to 1 volt (for example)….then each of the component signals that represent a single channel must have a peak amplitude of about 1 114 Copyright 2001, S.D. Personick. All rights volt/[75**(1/2)]~0.1V Cable TV FDM: Issues • Distortion: If the transmitter and/or the line amplifiers are not linear, then we can end up with square law and higher order terms that result in interference among the FDM channels • Noise; and interference from signals leaking into the cable system 115 Copyright 2001, S.D. Personick. All rights Impact of Distortion on FDM • Distortion (simplified example): s(t) is an FDM composite signal (e.g., 6 MHz TV channels stacked up in frequency) r(t) = s(t) + b [s(t)]**2 + c[s(t)]**3 116 Copyright 2001, S.D. Personick. All rights Impact of Distortion (cont’d) • If s(t) contains components centered at around 50MHz (for example) then b[s(t)**2] will contain frequencies at around 100MHz…interfering with the components of s(t) at 100MHz • If s(t) contains components centered at 100MHz and at 150MHz, then b[s(t)**3] will contain frequencies at 50 MHz, 200 MHz, 300MHz, 350 MHz, 400 MHz, and 450 MHz…causing interference at 117 all of thoseCopyright frequencies 2001, S.D. Personick. All rights Impact of Noise • Thermal noise associated with the finite temperatures of the cable sections, plus noise associated with the line amplifiers and the receivers connected to the system set a lower bound or baseline of system noise. This noise tends to appear as “snow” on a television receiver • “Ingress noise” (interference) from radio signals, electric motors, etc, represent 118 the more serious problem in most cable Copyright 2001, S.D. Personick. All rights Radio Frequency Interference • Another consideration, of importance in all metallic cable systems, is the possibility that the cable system will radiate signals that may interfere with other communications or information systems/appliances (e.g., medical devices in a hospital environment) • Radiation can occur due to imperfect shielding on coaxial cables, bad connectors, unterminated splitter output 119 ports, and Copyright imbalances in wire pair cable 2001, S.D. Personick. All rights “Wiring” Concerns Unused output port not terminated properly Wrong kind of “wire” Poorly designed splitter Bad or loose connector 120 Copyright 2001, S.D. Personick. All rights Personick’s ad-hoc equalizer 100 MHz = 628 M radians per second => 50 C = 1/[6.28 x 10**8] ~ 3.2 pf 50 ohms C 50 ohms 10 ohms 121 Copyright 2001, S.D. Personick. All rights Telecommunications Networking I Topic 5 Quantifying the Performance of Digital Communication Systems Dr. Stewart D. Personick Drexel University 122 Copyright 2001, S.D. Personick. All rights Digital Communication System Communication link Data input: 1011101 Data output: 1011101 123 Copyright 2001, S.D. Personick. All rights Point-to-Point Digital Communications: The Concept of “Errors” 1 0 1 0 1 1 time Input information: …1 0 1 0 1 1 ... 1 0 1 0 0 1 Output information: …1 0 1 0 01 124 Copyright 2001, S.D. Personick. All rights What Causes Errors? • Errors are caused by noise, distortion, and interference in the communications link, and can also be caused by “synchronization” errors and “congestion” • We can have two types of errors: - “miss” : we mistake a “1” for a “0” - “false alarm” : we mistake a “0” for a “1” 125 Copyright 2001, S.D. Personick. All rights Tradeoffs Between Misses and False Alarms • We can reduce the “miss” probability to zero by declaring that each output bit value will be a “1”; but then we will get a “false alarm” every time the true value of the corresponding input bit is a “0” • We can reduce the “false alarm” probability to zero by declaring that each output bit value will be a “0”; but then we will get a “miss” every time the126 Copyright 2001, S.D. Personick. All rights Probability of a “False Alarm” Trading Off “Miss” and “False Alarm” Errors 0 1 Receiver Operating Characteristic 1 Probability of a “Miss” 127 Copyright 2001, S.D. Personick. All rights Probability of a “False Alarm” Trading Off “Miss” and “False Alarm” Errors 0 1 Blended receiver Operating Characteristic 1 Probability of a “Miss” 128 Copyright 2001, S.D. Personick. All rights Classical Additive Noise Detection Problem + r(t) s(t) = 1 volt peak n(t)= white, Gaussian noise 129 Copyright 2001, S.D. Personick. All rights Matched Filter Sample here r=s+n h(t)=a(-t) a(t)=pulse shape Matched Filter 130 Copyright 2001, S.D. Personick. All rights Matched Filter Output: r r = s + n (volts) “s” is equal to either: S or 0 (volts); corresponding to a digital “1” or a digital “0” n is a Gaussian random variable with variance N (volts**2) 131 Copyright 2001, S.D. Personick. All rights Matched Filter Output: r S volts r= s + n Decision Threshold = D volts If r > D, then we will guess that s=S (I.e., that we have a “1”. If r < D, then we will guess that s=0 (I.e., that we have a “0” 0 volts 132 Copyright 2001, S.D. Personick. All rights Calculating the Error Probabilities The probability of a miss is given by: ERFC* [ (S-D)/(N**0.5)]; where ERFC* [ ] is called the “error function complement star” For example ERFC*[6] = 10**-9 0 D S 133 Copyright 2001, S.D. Personick. All rights Gaussian Shape Standard deviation 134 Copyright 2001, S.D. Personick. All rights Calculating ERFC*(x) (“earf-see-star”); also known as the Q function: Q(x) ERFC* (x) = the integral from x to infinity of: {1/[(2 pi)**(0.5)]} {exp -(y**2)/2} dy = Q(x) 135 Copyright 2001, S.D. Personick. All rights Example • We have a point-to-point metallic cable system • r(t)= s(t) + n(t) (volts) ; where: s(t) is a sequence of pulses, modulated on or off, where the pulse shape is a(t) n(t) is white Gaussian noise with spectral density, N, equal to 2kTR (volts**2/Hz) Copyright 2001, S.D. Personick. All rights 136 Example (cont’d) • R is the impedance of the metallic cable over which s(t) is being communicated; kT = 4 x 10**-21 Joules • We will assume that the energy in a single received pulse a(t) = integral of {[a(t)]**2 dt}/R = E Joules • If we pass r(t) through a matched filter with impulse response h(t) = C a(-t), 137 where C is aCopyright constant, then … 2001, S.D. Personick. All rights Example (cont’d) … The output of the matched filter will be r = s + n; where: • s= S or 0 volts; n is a Gaussian random variable with variance N= 2kTR(C**2)ER (volts**2); and S = CER (volts) • The ratio of (S/2)/(N**0.5)= [E/(8kT]**0.5 • If we wish to have an error rate of 10**-138 Copyright 2001, S.D. Personick. All rights Example (cont’d) • … we require: [E/(8kT)] = 36 or more • Since 8kT ~ 3.2 x 10**-20, we require the received pulse energy E to equal ~ 10**18 (Joules) or more. • If the transmitted pulse is 1 volt peak, and 100ns in duration; and the cable has an impedance of 50 ohms, then the transmitted pulse energy is: .02 x 10**-7 = 2 x 10**-9 (Joules) 139 Copyright 2001, S.D. Personick. All rights Telecommunications Networking I Topic 6 Point-To-Point Digital Communications Dr. Stewart D. Personick Drexel University 140 Copyright 2001, S.D. Personick. All rights Digital Point to Point Communications • In a real digital communication system, one has to be concerned with noise, interference and other effects that can cause errors • In the previous discussion, we briefly covered the topic of additive noise, and its impact on errors (misses and false alarms) • In metallic cable systems, “intersymbol 141 interference” is a key factor that we Copyright 2001, S.D. Personick. All rights Intersymbol Interference Transmitted pulse stream T Cable output (dispersion) + T 142 Copyright 2001, S.D. Personick. All rights Intersymbol Interference: Equalization Cable: H(f) Equalizer: 1/[H(f)] 143 Copyright 2001, S.D. Personick. All rights Digital Regenerator (Repeater) Cable: H(f) Equalizer & Matched Filter Decision circuit Clock signal Timing Recovery 144 Copyright 2001, S.D. Personick. All rights Timing recovery Pulse stream Clock (x)**2 Filter or PLL PLL=Phase-locked Loop 145 Copyright 2001, S.D. Personick. All rights Decision Circuit Pulse stream Threshold Compari tor D Flip Flop Clock 146 Copyright 2001, S.D. Personick. All rights Eye Diagram 147 Copyright 2001, S.D. Personick. All rights Eye Diagram (cont’d) 148 Copyright 2001, S.D. Personick. All rights T-Carrier • First introduced by the former Bell System in 1962. The first digital transmission system • Digital transmission rate is 1.544 Mbps • Works on 24 gauge wire pair cables • Repeaters every 6000 ft ~ 2km • Max cable loss at 772MHz ~35 dB • The signal is a “DS1” (but everyone 149 calls it a “T1”) Copyright 2001, S.D. Personick. All rights T-Carrier (cont’d) • Maximum cable length between repeaters limited by crosstalk: interference from other signals on other pairs in the same cable • Transmitted signal format: + or - 3V equals a logical “1”; 0V equals a logical “0”, no more than 7 “zeros” in a row are permitted • 00000000 is mapped to 00000001 originally; recently mapped to ++0000-150 (“B8ZS”) Copyright 2001, S.D. Personick. All rights Analog-to-Digital Conversion (and vice versa) Sampling Theorem: If we sample an analog signal at twice its highest frequency, we can reproduce it exactly from its samples 151 Copyright 2001, S.D. Personick. All rights A/D Conversion Example: Voice signals… -Highest frequency is limited (by a filter) to 4kHz -We sample this band limited signal at 8000 samples per second (125 microseconds between samples) -We represent each sample with 1 byte (positive and 152 Copyright 2001, S.D. Personick. All rights negative values are both captured by 256 D/A Conversion Samples reconstructed from the received digital bit stream Reconstructed waveform filter 153 Copyright 2001, S.D. Personick. All rights Multiplexing 24 inputs, each at 64 kbps T1 multiplexer; also known as a “D-channel bank” 1 output at 1.544 Mbps [24 x 64 kbps] + 8kbps = 1.544 Mbps; 8 kbps = overhead Output signal is a “DS1”, but everyone calls it a “T1” 154 Copyright 2001, S.D. Personick. All rights D-Channel Bank Frame Format 1st data byte, 2nd data byte, ……, 24th data byte, F 125 microseconds F= Framing bit bits Frame length is (8 x 24) + 1 = 193 8000 frames per second, corresponding to the rate at which voice signals are sampled 193 bits per frame x 8000 frames per second = 1.544 155 Mbps Copyright 2001, S.D. Personick. All rights Multiplexing Standards • In the United states, some common multiplex standards are: (above DS1, they are used with radio or fiber optic transmission systems) -DS1 (called T1) 1.544 Mbps -DS3 (called T3) 44.7 Mbps (28 DS1 + overhead) -STS1 (SONET-1) 51.84 Mbps -STS3 (SONET-3) 155.52 Mbps -STS 12 622.08 Mbps -STS 48 2.48832 Gbps -STS 192 ~10 Gbps 156 Copyright 2001, S.D. Personick. All rights Other Popular Digital Metallic Cable Transmission Systems • Ethernet: 10Base-T, 100Base-T; coax versions • Telephone modems: up to 56 kbps; limited by end-to-end switched telephone network • ADSL: asymmetric digital subscriber line; e.g., 1.5Mbps downstream, 384 kbps upstream …using only the existing telephone loop (not the switched network) • Cable modems: ~ 20 Mbps downstream, shared with other users; upstream depends 157 Copyright 2001, S.D. Personick. All rights Ethernet: 10Base-T Twisted Pair NIC Hub Computer NIC= Network Interface Card To other hubs or router 158 Copyright 2001, S.D. Personick. All rights Modem Subscriber loop A/D modem Computer (Philadelphia) D/A Public Switched Telephone Network (PSTN) modem Computer (Los Angeles) 159 Copyright 2001, S.D. Personick. All rights ADSL Loop pair ADSL Switched voice DSLAM computer PSTN Router Packet data To the Internet DSLAM=Digital subscriber line access multiplexer 160 Copyright 2001, S.D. Personick. All rights Cable Modem Splitter TV Coaxial Cable Coaxial Cable Twisted pair Modem PC 161 Copyright 2001, S.D. Personick. All rights Telecommunications Networking I Topic 7 Fiber Optic Transmission Systems Dr. Stewart D. Personick Drexel University 162 Copyright 2001, S.D. Personick. All rights Fiber Optics: Overview • 1966 C. Kao et. al, propose that strands of glass can be produced, which can carry light over long distances (>2 km) • 1970 First demonstration of a fiber with less than 20dB/km of loss (Maurer, et.al., at Corning) • 1975-77 Experiments and field trials • 1979 Real systems are placed in 163 Copyright 2001, S.D. Personick. All rights Basic Fiber Optic Transmission System Digital pulses (On/Off) Light Source Detector/Receiver Glass fiber Optical Transmitter Digital pulses 164 Copyright 2001, S.D. Personick. All rights The Radio Spectrum • AM Radio ~ 1 MHz (300 meter wavelength) • Television ~ 50-500 MHz • Digital cordless phone ~ 900 MHz • Wireless LAN ~ 2.5 - 5 GHz • DBS ~ 10 GHz (0.3 meter wavelength) • Fiber optics ~ 0.9 - 1.55 um (not visible) • Visible light ~ 4-7.5 x 10**14 Hz (0.8-0.4 um) 165 Copyright 2001, S.D. Personick. All rights Optical Transmitter: example Data in 2V Peak ~20 mA peak Light output 50 ohm resistor Light Emitting Diode 166 Copyright 2001, S.D. Personick. All rights Optical Transmitter Example driver bias Laser Light output Data in 1V peak current Light output current Copyright 2001, S.D. Personick. All rights 167 Simplified Semiconductor Injection Laser Curren t 168 Copyright 2001, S.D. Personick. All rights Optical Transmitter: example • Current = 20mA = .020A • # electrons per second = .020 A/[1.6 x 10**-19] Coulombs per electron = n • # photons produced/second =n x [Quantum Efficiency] • optical power out = n x QE x [~1.5 x 10**19 Joules per photon] ~ .020 QE x [1.5/1.6] (W) ~20 x [1.5/1.6] x QE (mW) 169 S.D. Personick. All rights • If QE~ 20%,Copyright then2001,power out ~3.75 Optical Fiber Optical Pulses Optical Pulses Fiber Optical output pulses are attenuated and spread in time compared to optical input pulses 170 Copyright 2001, S.D. Personick. All rights Causes of Attenuation • Light is absorbed by fiber impurities and the principal fiber material itself • Light is “scattered” out of the fiber because of the inherently random density fluctuations of any “glass” (Rayleigh scattering) as well as by more macroscopic density fluctuations • Typical long distance fiber: <0.5 dB per km Typical plastic fiber: >100 dB per km 171 Copyright 2001, S.D. Personick. All rights Causes of Pulse Spreading: Modal Delay Spread T (min) = nL/c, where c/n = speed of light in fiber T (max) = T(min) x [1/cos(max angle that is captured)] core cladding “Multimode” Fiber c =300,000,000 m/s, n~1.5... n/c ~ 5ns/m 172 Copyright 2001, S.D. Personick. All rights Modal Delay Spread •The rays in the previous slide represent the solutions of Maxwell’s equations…each of which is called a “mode” •If one actually solves Maxwell’s equations, one finds a discrete set of modes, each corresponding to a ray at a different angle D relative to the axis Dspacing between these allowed rays is •The •If is large enough (e.g., 0.2 radians corresponding to ~11.4 degrees, then only the axial ray is below the critical angle. 173 Copyright 2001, S.D. Personick. All rights Modal Delay Spread • If only the axial ray is below the critical angle, then there is only one solution to Maxwell’s equations (one mode) which is guided by the fiber. • Such a fiber is called a single mode fiber • With only one ray (mode) there is no modal pulse spreading! 174 Copyright 2001, S.D. Personick. All rights Dispersion: ps/km-nm Causes of Pulse Spreading Dispersion: a change in the delay down the fiber as the wavelength Zero dispersion at ~1.3 um changes- ps/[nm-km] Wavelength 175 Copyright 2001, S.D. Personick. All rights Causes of Pulse Spreading • Pulse spreading can be caused by the variation of delay vs. angle in multimode fibers (delay spreading). Typical plastic multimode fiber: >100 ns/km • Pulse spreading can also be caused by the variation of delay with wavelength (“dispersion”). Typical glass fiber with 900 nm LED source ~5 ns/km; with 1550 nm laser source < 0.1 ns/km 176 Copyright 2001, S.D. Personick. All rights Pulse Spreading: Examples • Multimode fiber: Maximum angle captured in fiber is 0.2 radians (for example)~11.5 degrees. 1/cos(0.2 rad) = 1.020. Delay spreading = 5 ns/m x 0.02 = 0.1 ns/m = 100 ns/km • Single mode fiber +900 nm LED source: Dispersion at 900 nm wavelength ~100 ps/nm-km. LED spectral width ~50 nm. Dispersion ~ 100 x 50 = 5000ps/km 177 =5ns/km Copyright 2001, S.D. Personick. All rights Optical Receiver Amplifier + Regenerator Output pulses Photodiode 178 Copyright 2001, S.D. Personick. All rights Optical Receiver • The detector converts photons to electrons ~ 0.5 mA/mW (output current/input power) • The amplifier amplifies the weak current that is produced by the detector in a typical optical fiber application • The regenerator produces a new 179 electrical pulse stream (clock recovery, Copyright 2001, S.D. Personick. All rights Causes of Errors in Optical Fiber Systems • Noise produced by the amplifier in the receiver • “Quantum” noise associated with the detection process • Intersymbol interference due to pulse spreading • Bottom line: In a typical fiber optic system, we require ~20,000 received photons per pulse to produce an error rate of 10**-9; assuming that we don’t have a significant 180 amount of intersymbol interference (pulse Copyright 2001, S.D. Personick. All rights Fiber Optic System: example Digital pulses (On/Off) Light source Detector/Receiver Glass fiber Optical Transmitter Assume: Bit rate = 100Mbps; Optical transmitter output = 1 mW; Coupling loss into fiber = 3dB; Pulse spreading<0.1 ns/km; Fiber loss = 0.5 db/km; Required optical energy per received pulse: 20,000 photons x 1.5 x 10**-19 J/photon Digital pulses 181 Copyright 2001, S.D. Personick. All rights Optical Fiber System: example • Receiver requires 20,000 photons per received optical pulse = 20,000 x 1.5 x 10**19 J per pulse = 3 x 10**-15 J/pulse • Bit (pulse) rate is 100Mbps; therefore the average received power level must be greater than 0.5 x (3 x 10**-15) x (10**8)= 1.5 x 10**7 watts = 1.5 x 10**-4 mW~ -38.2 dBm 182 Copyright 2001, S.D. Personick. All rights Optical Fiber System: example • Transmitter average power into the fiber is: 1 mW x 0.5 (coupling loss) x 0.5 (duty cycle) = 0.25 mW • Allowable loss = 0.25/0.00015 = 1.66 x 10**3 ~ 32.2 dB • At 0.5 dB/km loss, we can allow ~64 km of fiber [44 km] • Checking the pulse spreading; we get: 64 km x 0.1 ns/km = 6.4 ns (0.64 x pulse spacing) • Bottom line, if we allow up to 50 km of fiber, we will be within the pulse spreading limit, and we will have about 14 x 0.5 = 7 dB of margin w.r.t. noise 183 limited operation [pulse spreading limit is 5 km] Copyright 2001, S.D. Personick. All rights Telecommunications Networking I Topic 8 Wireless Transmission Systems Dr. Stewart D. Personick Drexel University 184 Copyright 2001, S.D. Personick. All rights Wireless Point-to-Point Link Antenna Feed Line (e.g., coaxial cable) Radio Transmitter Radio Receiver 185 Copyright 2001, S.D. Personick. All rights The Electromagnetic Spectrum • 30-300Hz: SLF 3GHz-30GHz: SHF (DBS) • 300Hz - 3kHz ULF 30GHz-300GHz: EHF • 3kHz - 30kHz: VLF 300,000GHz: 1um light • 30kHz-300kHz: LF • 300kHz-3MHz: MF (AM Radio) • 3MHz-30MHz: HF (Short Wave Radio) • 30MHz-300MHz: VHF (FM, TV) • 300MHz-3GHz: UHF (TV, Digital Cordless, 186 Copyright 2001, S.D. Personick. All rights …) Wireless Transmitter Antenna Informatio n Modulator IF Oscillator Mixer RF Amp RF Oscillator 187 Copyright 2001, S.D. Personick. All rights Transmitter Subsystems • 1 milliwatt -100 milliwatts: low r.f. exposure with hand held appliances, low battery drain (some hand held appliances radiate ~5 watts…but I wouldn’t hold one of these near my head!) • 10 watts - 100 watts: okay for consumer and small business applications, with 115 volt or 12 volt automobile power. Stay 10-30 feet from the antenna. 188 Copyright 2001, S.D. Personick. All rights Transmitter Subsystems R.F. Power = (V**2)/Z Antenna V~ 1 volt Z~50 ohms FET Coax R.F. Amplifier R.F. Power (launched into the coaxial cable) = 1/50 Watt = 20 mW R.F. Power radiated = ? 189 Copyright 2001, S.D. Personick. All rights The Antenna Reflected Power Radiated Power Forward Power 190 Copyright 2001, S.D. Personick. All rights The Antenna • If the antenna is much smaller in length than the wavelength of the radiation (c/f), then the antenna will be a very inefficient radiator (most of the forward power is reflected) • If the antenna is 1/4 wavelength or larger in size, it can be an efficient radiator • If the antenna is significantly larger in size than 1 wavelength, it can be an efficient and directional radiator 191 Copyright 2001, S.D. Personick. All rights A Directional Radiator Reverse: fields cancel 1/4 wavelength spacing between two dipoles. 3/4 wavelength delay in crossover cable Forward: fields add It’s actually not quite that simple: dipole interactions 192 Copyright 2001, S.D. Personick. All rights The Parabolic Dish Antenna Divergence angle ~Lambda/D 193 Copyright 2001, S.D. Personick. All rights Wireless Receiver Antenna Informatio n Demodulator IF Oscillator Mixer RF Amp RF Oscillator 194 Copyright 2001, S.D. Personick. All rights The Receiver Subsystem Signal + Background Noise + Interference Cable attenuation + thermal noise + Amplifier noise Low Noise Amplifier 195 Copyright 2001, S.D. Personick. All rights Sources of Thermal Noise Sun Background (27K) Signal Field of View Mirror Hot object (333K) 196 Copyright 2001, S.D. Personick. All rights The Receiver Subsystem Example: Equivalent background temperature: T = 100K (Kelvins) Equivalent background noise = kTB (watts) Coupling loss of antenna into cable ~ 0dB Cable loss = 3 dB Cable temperature = 293K Preamplifier Noise Temperature = 30K Equivalent amplifier total input noise = ??? Copyright 2001, S.D. Personick. All rights 197 The Receiver Subsystem Example (continued): Equivalent input noise = kB [100/2 + 293/2 + 30]=226.5kB I.e., half of the background noise + half of the cable noise + the preamplifier noise 198 Copyright 2001, S.D. Personick. All rights The End-To-End System Example (continued from prior example): Assume that the transmitter power amplifier produces 20 milliwatts. The coupling loss from the transmitter into its coaxial antenna feed cable is 0 dB. The transmitter antenna feed cable has 1.5 dB of loss. The antenna radiates 90% of the power that arrives from the transmitter antenna feed cable, and reflects the rest back into the cable. How much power is radiated? Assume that the bandwidth, B = 6 MHz. How much total propagation loss can we 199 allow if the required signal-to-noise ratio is 40 dB? Copyright 2001, S.D. Personick. All rights The End-to-End System Example: (continued) The transmitter produces 20 mW = +13 dBm The loss of the cable, plus the impact of a 10% reflection (90% radiation) is: 1.5 dB (cable loss) - 10 log (0.9) = 1.95 dB The total radiated power = +11.05 dBm = 12.7 mW The required power at the preamplifier input = 226.5 x k x 6 x 10**6 x 10**4 (watts)... I.e., 40 dB larger than the noise 200 Copyright 2001, S.D. Personick. All rights The End-to-End System • The radiated power = 12.7 mW • The required power at the receiver preamplifier input = 1.85 x 10**-7 mW • The required power at the receiver antenna is 2 x 1.85 x 10**-7 mW = 3.7 x 10**-7 mW • The allowed propagation loss is the ratio of these ~3.4 x 10**7 ~ 75.3 dB 201 Copyright 2001, S.D. Personick. All rights Calculating the Propagation Loss Antenna equivalent area = A Area of surface = 4 r**2 Copyright 2001, S.D. Personick. All rights 202 Calculating the Propagation Loss Suppose the frequency is 100 MHz, and the wavelength = 3 meters. Assume that the equivalent area of the antenna is 2.25 square meters. If the allowable loss is 75.3 dB, then the distance from the transmitter to the receiver, r , can be derived from: 2.25/[4 r**2) > 1/[3.4 x 10**7]; r < 2.5 km (Line-Of-Sight) 203 Copyright 2001, S.D. Personick. All rights Why Do Broadcast Transmitters Have Such High Power? In the previous example, we could cover 2.5 km with 20 mW. If we want to cover a 100 km radius, we would need 40**2 or 1600 times the power. That would correspond to 32 watts << 10 kW But, what about: high antenna-feeder-cable losses, splitters, noisy preamplifiers, inadequate antennas (low effective area, and poor matching to the feeder cable), 204 attenuation through buildings, and fading Copyright 2001, S.D. Personick. All rights Adaptive Antennas Combiner Multiple input antennas Jamming signal Desired input signal Coherently combined output signal Copyright 2001, S.D. Personick. All rights 205 Adaptive Antennas • An adaptive antenna can combine the signals received by a multiplicity of component antenna (in various amplitude and phase relationships) to null out one or more jamming signals 206 Copyright 2001, S.D. Personick. All rights 207 Copyright 2001, S.D. Personick. All rights 208 Copyright 2001, S.D. Personick. All rights Lasercom 209 Copyright 2001, S.D. Personick. All rights Telescopes Receiver Transmitter 210 Copyright 2001, S.D. Personick. All rights Telescope Receiver Transmitter 211 Copyright 2001, S.D. Personick. All rights T R 212 Copyright 2001, S.D. Personick. All rights R T Pointing errors and beam wander due to clear air turbulence 213 Copyright 2001, S.D. Personick. All rights Bio-Complexity Network Van Rensselear Hall (33rd St) Gbps Ethernet [fiber] Light beams Telescope/Optical Transceiver Remote observation and control (Dr. Banu Onaral et. al). Commonwealth Hall (31st St) Gbps Ethernet [fiber] Cellular Observatory Microscope System (Dr. J.Yasha Kresh et. al.) MCP Hahnemann (15th St) 214 Copyright 2001, S.D. Personick. All rights cey 215 Copyright 2001, S.D. Personick. All rights 216 Copyright 2001, S.D. Personick. All rights Telecommunications Networking I Topic 9 Switching Fundamentals Dr. Stewart D. Personick Drexel University 217 Copyright 2001, S.D. Personick. All rights Switching Fundamentals The Classical Switching Opportunity # of links: N(N-1)/2 End system End system e.g., Telephone 218 Copyright 2001, S.D. Personick. All rights The Classical Switching Opportunity # of links: N Switch 219 Copyright 2001, S.D. Personick. All rights What’s a Switch? Crossbar switch Signal path closed crosspoint wires wires # of crosspoints: N 2 220 Copyright 2001, S.D. Personick. All rights Mechanical Crosspoint: Reed Relay Apply a current to the coil, and the resulting magnetic field forces the reeds to touch Coil Reeds Glass tube Control Signal 221 Copyright 2001, S.D. Personick. All rights Multi-stage Switching Fabric # of cross points: 2N ~ 3/ 2 1 1 9 3x3 #1 3x3 #1 3x3 #2 3x3 #2 3x3 #3 3x3 #3 9 9x9 crosspoints vs. 9x6 cross points (with blocking) 222 Copyright 2001, S.D. Personick. All rights Blocking Blocking occurs when a input cannot reach an idle output Previous example: Blocking on group 1 occurs if: 2 or 3 inputs try to reach the same output group Probability of blocking = [1/3 + 1/3] - 1/9 =5/9 Switching theory: how to design switches with an acceptable # of cross points, and an acceptable level of blocking 223 Copyright 2001, S.D. Personick. All rights Banyan Switching Fabric # of cross points =2 N log 2 1 N 1 2x2 8 8 224 Copyright 2001, S.D. Personick. All rights Time Division Switching Channel bank Time-Space-Time Digital Switch 7 TSI Space Switch TSI 19 TSI= Time Slot Interchanger Input and output streams are time-division multiplexed 225 Copyright 2001, S.D. Personick. All rights There’s More to Switching than Switches • For example: -On-hook/Off-hook detection -Providing dial tone -Signaling (e.g., tones or dial pulses) -Call processing -Advanced services (e.g., caller ID) -Routing (finding a path to the destination) -Automatic Message Accounting (AMA)226 Copyright 2001, S.D. Personick. All rights On-Hook/Off-Hook Detection Line Card in Central Office Telephone Line Interface Circuit inductor Hook switch Specification: “Off hook” = 20-120 ma of “loop” current 227 Copyright 2001, S.D. Personick. All rights Signaling a b d 1 2 e 4 5 7 8 * 0 c 3 Send two tones… Choose: 1 of {a,b,c} and 1 of {d,e,f,g} Total of 12 combinations 6 f DTMF*=Dual tone multifrequency 9 g # *Also known as: TouchTone(R) 228 Copyright 2001, S.D. Personick. All rights Signaling STP SS7 Switch Switch Trunk Group STP=Signaling Transfer Point SS7= Signaling System Number 7 229 Copyright 2001, S.D. Personick. All rights Finding a route to the Destination • Telephone switches are uploaded with “translation tables” which map dialed telephone numbers to predetermined output ports on the switch. Thus, when I dial 1-215-895-6208 from home in NJ, my local switch “translates” the first six numbers (plus the “1”), and knows that this is a call that must be passed to an “access tandem” switch…for routing on230 Copyright 2001, S.D. Personick. All rights to my chosen long distance carrier. Automatic Message Accounting • Telephone switches collect “automatic message accounting” (AMA) information: called number, calling number, start time, end time, special features used; and store it on a disk. This information is periodically downloaded by a billing center for offline processing into telephone bills. A telephone company’s billing system is a strategic competitive asset. 231 Copyright 2001, S.D. Personick. All rights Telecommunications Networking I Topic 10 Telecommunications Network Management Dr. Stewart D. Personick Drexel University 232 Copyright 2001, S.D. Personick. All rights Telecommunications Network Management • Some questions to ask if you are the owner/operator of a large telecommunications network: -How do I characterize the quality of service I will provide in a way that can be measured and engineered into my networks; and in a way that my customers will find useful Copyright 2001, S.D. Personick. All rights 233 Telecommunications Network Management • Some questions to ask if you are the owner/operator of a large telecommunications network: -How will I ensure that I have enough network capacity to provide my customers with a specified, quantifiable quality of service…but not more network capacity than what is required to do so 234 Copyright 2001, S.D. Personick. All rights Telecommunications Network Management • Some questions to ask if you are the owner/operator of a large telecommunications network: -How do I activate services for customers without having to send technicians out to the customers’ physical location (service activation) -How do I monitor each customer’s Copyright 2001, S.D. Personick. All rights quality of service (service assurance) 235 Telecommunications Network Management • Some questions to ask if you are the owner/operator of a large telecommunications network: -How to I monitor the current state (configuration, alarm conditions, and usage measurement data) of my network equipment; and how can my network automatically recover from Copyright 2001, S.D. Personick. All rights various fault conditions 236 Operations Support Systems (OSS’s) • Operations Support Systems (OSS’s), not to be confused with operating systems, are complex computer applications which are used to automate many of the tasks that were done manually, in telecommunications systems, a few decades ago. • The objectives are to increase responsiveness to customers needs, and to reduce cost • Similar to networked information systems in 237 Copyright 2001, S.D. Personick. All rights other industries Telecommunications Management Networks (TMN) Security Enterprise Management Systems Billing Service Management Systems Traffic Mgmt/QOS Connectivity Network Management Systems Redundancy Routing Element Management Systems ... Network Elements 238 Copyright 2001, S.D. Personick. All rights A Classic Traffic Engineering Problem Concentrator 1 1 Lines N Trunks M<N Erlang’s formulas (queuing theory 239 Copyright 2001, S.D. Personick. All rights The Poisson (Random) Process Events x x Time x x x x The probability of an “event” in any small is dt, where “ ” is called interval of length dt (seconds) the intensity of the Poisson random process. The occurre of “events” in any interval is statistically independent of the occurrence of “events” in any other disjoint interval 240 Copyright 2001, S.D. Personick. All rights The Poisson (Random) Process Events x x x x x x The probability of N “events” in an interval of length T seconds is: e N Where is / N! T 241 Copyright 2001, S.D. Personick. All rights Example Suppose there are 100 telephone lines connected to a concentrator with M outgoing trunks. Each telephone line generates calls (events) approximately as a Poisson process, at an intensity of 3 calls per hour; and each call lasts exactly 3 minutes. (Clearly a simplification) How many outgoing trunks, N, do we need to ensure the probability of “all trunks busy” at any given time is less than .01? “All trunks busy” would occur if we had M or more “events” in the Copyright last 3 2001, minutes S.D. Personick. All rights 242 Example (Cont’d) Probability of “all trunks busy” = M Where N e / N! < .01 =300 x 3/60 = 15 Answer M = 26 243 Copyright 2001, S.D. Personick. All rights Intelligent Networks 244 Copyright 2001, S.D. Personick. All rights Conventional Telephone Network with SS7 STP Janet Switch STP Switch Bo b STP= Signaling Transfer Point 245 Copyright 2001, S.D. Personick. All rights Call Processing Bob’s Call State: Network Janet’s Call State: Time Idle (on hook) Active (off hook) Dial tone Dialing (digit collection) Post Dial (waiting) Active/Busy (talking) Idle (on hook) Idle (on hook) Ringing Active/Busy Idle (on hook) 246 Copyright 2001, S.D. Personick. All rights Intelligent network-based 800 Service 800 SMS 800 SCP STP Janet Switch STP Switch Bo b STP= Signaling Transfer Point SCP = Service Control Point SMS=Service Management System 247 Copyright 2001, S.D. Personick. All rights Call Processing Bob’s Call State: Network Janet’s Call State: Time Idle (on hook) Active (off hook) Dial tone Dialing (digit collection) Play announcement Digit Collection Post Dial (waiting) Active/Busy (talking) Idle (on hook) Idle (on hook) Ringing Active/Busy Idle (on hook) 248 Copyright 2001, S.D. Personick. All rights Service Creation Environment Non-real-time download SCE STP ISCP (data + call processing) Switch Intelligent Peripheral (IP) 249 Copyright 2001, S.D. Personick. All rights Advanced Services: Example “Calling Party Name” LIDB Query/response SW SW Bob Incoming call Janet Bob dials Janet. Bob’s switch queries the line information data base (LIDB) for the name associated with Bob’s line. This information is passed to Janet’s switch using SS7. Janet’s switch delivers this information between the first and second ring of Janet’s phone, using a modem-like signal. 250 Copyright 2001, S.D. Personick. All rights “Calling Party Name” Bob dials Janet. Bob’s switch queries the line information data base (LIDB) for the name associated with Bob’s line. This information is passed to Janet’s switch using SS7. Janet’s switch delivers this information between the first and second ring of Janet’s phone, using a modem-like signal. It is also possible to implement this functionality by having Janet’s switch send a query to Bob’s LIDB only if Janet subscribes to the calling name delivery service (rather than sending the information for every call). 251 Copyright 2001, S.D. Personick. All rights Service Creation Example • Customer IP off-hook Switch SCP Prov inst 252 Copyright 2001, S.D. Personick. All rights Service Creation Example • Customer IP Switch off-hook SCP Prov inst Play announcement #1 & Collect 5 digits 253 Copyright 2001, S.D. Personick. All rights Service Creation Example • Customer IP Switch off-hook SCP Prov inst Play announcement #1 & Collect 5 digits “Please enter you access pin number” 254 Copyright 2001, S.D. Personick. All rights Service Creation Example • Customer IP Switch off-hook SCP Prov inst Play announcement #1 & Collect 5 digits “Please enter you access pin number” 54321# Authenticate Authenticate 255 Copyright 2001, S.D. Personick. All rights Service Creation Example • Customer IP Switch off-hook SCP Prov inst Play announcement #1 & Collect 5 digits “Please enter you access pin number” 54321# Authenticate Authenticate OK Play announcement # 2256& Process call Copyright 2001, S.D. Personick. All rights Service Creation Example • Customer IP Switch off-hook SCP Prov inst Play announcement #1 & Collect 5 digits “Please enter you access pin number” 54321# Authenticate Authenticate OK Play announcement # 2257& Process call Copyright 2001, S.D. Personick. All rights Advanced Intelligent Network –based Mobility Management Janet’s HLR Janet’s current VLR Radio port controller Switch Switch Radio port Janet Bo b HLR=Home Location Register VLR=Visited Location Register 258 Copyright 2001, S.D. Personick. All rights Advanced Intelligent Network –Domino’s Pizza Time? Date? Bob’s number? Bob’s zip code? Load management Domino’s 973 829 1703 SCP Switch 2 Switch 1 Bo b 1. Bob (in area code 908) dials: 366-4667 (Dominos) 2. Switch 1 recognizes this as a number requiring “translation” 3. Switch 1 queries SCP. 4. SCP determines and returns the number of nearest open Domino’s: (973) 829-1703 Copyright 2001, S.D. Personick. All rights 259 Advanced Intelligent Network –Domino’s Pizza Call for fire support Time? Date? Bob’s ID? Bob’s location Load management Fire Support (nearest) Switch 2 SCP Switch 1 1. Bob (in location XYZ) originates a request to “fire support” 2. Switch 1 recognizes this as origination as requiring “translation” 3. Switch 1 queries SCP. 4. SCP determines and returns the network address of the nearest S.D. Personick. All rights available sourceCopyright of fire 2001, support Bo b 260