Department of Electrical and Computer
Engineering
Wireless Communications and Networking
ECE 250 -
Spring 2005
Prof. Volkan Rodoplu
Lectures: Monday/Wednesday
Office hours: Room 4113, Engineering I; Monday and Wednesday
(starting on
Midterm Exam:
Final Exam:
Announcements
|
1) Course grades have been submitted. Your course grade should be available on GOLD soon. 2) The final exam questions and their solutions will be
posted on this web page by 3) You may pick up your final exam if you drop by Prof. Rodoplu's office some time in the afternoons (this week). (Make arrangements by email beforehand.) 06/06/05: FINAL EXAM information: (1) The final will be a 3-HOUR in-class final exam. (Not 4 hours!). It will be 4:00-7:00 PM on June 10, 2005, in class. The final exam is open-book, and calculators are allowed (but NO laptops). (2) Sample Final Exams: There are no sample final exams. After the midterm: the topics covered were: DMT/OFDM systems, and lectures on sensor networks. The final exam material will include all the midterm material plus the DMT/OFDM topics (HW # 5) as well as the two papers assigned in class (HW # 6). For the DMT coverage, you may find it useful to take a look at any sample midterms/final exams at John Cioffi's web site (Stanford). 05/25/05: Homework # 6 has been posted on this page. 05/25/05: Homework # 5: Clarifications: Problem 1: The SNR_mfb = || P(1,:) ||^2 * E_x / (\sigma^2). That is, the SNR_mfb ("matched filter bound") is the channel norm squared times the average energy per symbol divided by the noise variance. Note that P(1,:) is the first row of the P matrix (hence, it is the discrete multipath delay profile vector). 05/18/05: HOMEWORK # 5: ALL STUDENTS must include the print-out of their MATLAB or Mathematica code, as well as the output files / lines showing their answers to problems that require MATLAB/Mathematica. (You may NOT share code; you have to write your own code.) 05/18/05: Clarifications on Homework # 5: Problem 1: (a) SNR_MFB refers to the "matched filter bound", namely the SNR performance attainable for a one-shot use of the channel, with a matched filter used at the receiver. For the AWGN channel, this is g * E / sigma^2; namely the channel gain times the energy per symbol divided by the noise power of AWGN. (b) For "loading algorithms": Use the waterfilling that we discussed in class. Problem 2: (a-b) Equation (10.149) is: y = P x + n This is the ISI channel equation that we wrote down to describe the channel for vector coding. y here is an N * 1 vector, and P is an N * (N + \nu) matrix. (c) Equation (10.155) is: \bar{H}_\bf{x} = N / 2 log_2(\pi e \abs{R_xx}^{1/N}) We did not cover this part in class (to show the optimality of vector coding). Please refer to John Cioffi's lecture notes online (Chapter 10 on multichannel modulation). (d) Equation (10.167) is: the definition of SNR_{vc}, namely the definition of the equivalent SNR for vector coding (which we covered in class). (d) Equation (10.16) is: \bar{b} = 1 / 2 log_2 ( 1 + SNR{m,u} / \Gamma) This equation says that the number of bits per dimension that can be transmitted is given by Shannon's formula, but where the SNR is taken as the multichannel SNR. (Please see Section 10.2.2 in John Cioffi's course reader). SNR_{m,u} is the multichannel SNR, namely the equivalent SNR defined for a set of parallel AWGN channels each of which has an SNR of SNR_{n}. 05/04/05: (1) Lecture 7 has been posted on this web page under Lecture Notes, (2) Sample Midterms # 1 - # 7 have been posted. The solutions to these midterms will be available under the Solutions link on May 6, 2005. 05/02/05: MIDTERM DATE HAS BEEN CHANGED TO MAY 11, 2005 (5:00-8:00 PM, in class). 04/27/05: OFFICE HOURS Office hours will take place 7:00-7:30 PM MW (after lecture), and the homework will be due FRIDAY 12:00 PM in Prof. Rodoplu's mailbox from now on. (This should give you time to ask questions before the homework is due.) 04/25/05: IMPORTANT ANNOUNCEMENTS: - MIDTERM date has been changed to May 9, 2005; Monday (in class) - Midterm will be comprehensive, in particular, it will include HW # 4 material. - The deadline for HW # 3 has been postponed to April 29, Friday, NOON, in the mailbox of Prof. Rodoplu (on 4th floor of Engineering I.) - HW # 4 will be posted on the web page on April 27, and will be due May 4, in class. (These changes have been noted on the homework deadlines on this web page.) 04/21/05: Homework # 3 has been posted on this web page. 04/21/05: Clarifications on Homework # 2: > 1. Are Ea, Eb, and Ec assumed to be equal before reflection? Yes. > 2. Should the plot of amplitude variation caused by multipath fading be > with respect to angle alpha? No. The angle theta is fixed here. We would like to see the amplitude variation plotted as a function of time (seconds). We would like to see how in time, the fading will be exhibited. Theta is indeed the acute angle in the picture. (So pi-theta should in fact be theta in that picture.) >On problem 5: > The question states "Write the expression for Phi(tau) when r(t) is the > signal received at isotropic scattering case for a mobile with constant > speed." > 1. Does isotropic scattering mean uniform scattering? Yes. > I answered this question by showing the following: > 1. phi of z = (phi of r (0))^2 + (phi of r(tau))^2 This is correct. > 2. For isotropic scattering I took the inverse Fourier transform of > your result for the power spectral density for uniform scattering. > Is this what you had in mind? This is good. You will get a Bessel function when you do this. (But if you write down the integral expression, that will be fine.) 04/18/05: Problem 2 on Homework # 2: Assume that interference and the large scale variation are independent. The final answer will be an integral (not in closed form). 04/18/05: Problem 4 on Homework # 2: Note that since the reflection coefficients are 0.5, the electric field intensities (amplitudes) are related as: E_B = E_C = E_A / 2. 04/11/05: Clarifications for Problem 2 on HW # 1: 1. The problem refers to Equations 2.3 and 2.5. These equations are the following equations from the lecture notes (Lecture 2): Equation 2.3: SIR = 1 / ( (1 / ( 2M + 1 )^n) + (1 / (2M - 1)^n) ) for downlink Equation 2.5: SIR = ( 2M - 1 )^n / 2 for uplink, close users. 2. For both parts (a) and (b) of this problem, assume that NO HANDOFF TAKES PLACE when the mobiles (on the boundaries) move into the adjacent cell. (So, do your calculations assuming that the interferer, for example, continues to use the same channel hence increasing the co-channel interference.) --- This makes sense because in practice, the handoff is not initiated based on the exact position of the cell boundary, but rather based on the SIR (signal-to-interference ratio); hence, for example, for the downlink in Part (a), we would initiate a handoff when the SIR falls below the given threshold (not when the mobile crosses the cell boundary, which is not directly measurable.) 04/11/05: All homeworks from now on will be due on Wednesdays (instead of Monday). Note that the due dates have been updated below (under Homework). 04/11/05: HW # 1 is due at the beginning of class on Wednesday (05/13/05). |
|
|
Homework
- Solutions
|
(due April 13, 2005) |
(due April 20, 2005) |
|
(due April 29, 2005) |
(due May 6, 2005) |
|
(due May 27, 2005) |
|
|
(due June 3, 2005) |
|
Sample Exams
Acknowledgments: I would like to thank Prof. Ted Rappaport, Prof. Andrea Goldsmith and Prof. Narayan Mandayam for making their lecture notes available online. I would also like to thank Prof. Donald Cox on whose lecture notes and assignments at Stanford some of this material is based.
Lecture Notes
|
·
Lecture 11 [see Sample Midterm # 1] ·
Lecture 12 [see Sample Midterms # 2
- # 7] |
|
|
|
|
|
|
|
|