Advanced Computer Networks Exercise Session 4. Qin Yin Spring Semester 2013

Transcription

1 Advanced Computer Networks Exercise Session 4 Qin Yin Spring Semester

2 Administration If you haven't received any about your submission We got your solutions for A1 & A2 About solutions Use your own words, minimize copy-paste Be precise about your points 2

3 COPE: Wireless Network Coding Consider the following scenario: 3

4 COPE Approach Increased throughput: saved transmission can be used to send more data 4

5 COPE Snooping Every node snoops on all packets A node stores all heard packets for a limited time Node sends Reception Reports to tell its neighbors what packets it heard Reports are piggybacked on packets If no packets to send, periodically send reports 5

6 COPE Coding To send packet p to neighbor A, XOR p with packets already known to A Thus, A can decode But how can multiple neighbors benefit from a single transmission? 6

7 Question 1 What are the assumptions that COPE makes? What are the implementation related tricks that COPE plays? 7

8 Basic loss estimation and rate change algorithm From Robust Rate Adaptation for Wireless Networks, Mobicom'06: Measure loss ratio over a window of packets Adjust the data rate accordingly 8

9 Question 2 In the paper, the author also proposed an improved algorithm with adaptive RTS filter. Explain briefly: how that algorithm works? how it improves over the basic algorithm? 9

10 Beacon transmission in a busy infrastructure network 10

11 Beacon transmission in a busy ad-hoc network 11

12 Power Management in

13 Power Management in (cont.) access points periodically send a beacon with a timestamp to synchronize all stations Power management implemented using beacons: beacons may contain a TIM or DTIM TIM = Traffic Indication Map, announces a set of unicast packet transmissions Contains the list of destinations of the buffered packets DTIM announces broadcast packet Stations always awake to receive 'T' and 'D' Stations stay awake if they are listed as receiver in 'D' or 'T' packet 13

14 Question 3 You've seen power management in IEEE infrastructure networks Explain why this mechanism doesn't work for adhoc networks. Propose a power management solution for IEEE ad-hoc networks. 14

15 The Advent of Cellular Networks Prior to cellular, mobile radio telephone system was based on: High power/transmitter/receivers Could support about 25 channels In a radius of 80 km To increase the network capacity Multiple low-power transmitters (100W or less) Small transmission radius area split in cells Each cell with its own frequencies The same frequency can be reused at sufficient distance 15

16 Question 4 Why do mobile network providers install several thousands of base stations which is quite expensive, throughout a country, and do not use powerful transmitters with huge cells like for example radio stations use? 16

17 The Hexagonal Pattern Used as an approximation for a cell Equidistant access to neighboring cells Center to center distance d = 3R, center to corner radius R Each cell features on base station Through power control cover the cell area while limiting power leaking to co-frequency cells Frequency re-use not possible in direct adjacent cells 17

18 Minimum cell separation D = minimum distance between centers of cochannels cells N = # of cells in a repetitious pattern, i.e. reuse factor N = 4 N = 7 18

19 System capacity and interference S = Total # of duplex channels available for use K = total # of duplex channels per cell S = k*n If a cluster is replicated M times within a system, the total # of duplex channels C can be used as a measure of capacity C = MkN = MS Tradeoff: If N decreases k increases (since S is a constant) If N decreases M increases (for a fixed geographical area) M increases system capacity C increases Price: D/R decreases co-channel interference increases 19

20 Cell splitting Smaller cells have greater system capacity Better spatial reuse As traffic load grows, larger cells could split into smaller cells Requires careful power control and possibly more frequent handoffs for mobile stations 20

21 Question 5 Consider a cellular system in which total available voice channels to handle the traffic are 960. The area of each cell is 6 km2 and the total coverage area of the system is 2000 km2 Calculate the system capacity if the cluster size, N (reuse factor), is 4. Do the same with N = 7 How many times would a cluster size 4 have to be replicated to cover the entire cellular area Does decreasing the reuse factor N increase the system capacity? Explain. 21

22 GSM Architecture MS: Mobile Station BTS: Base Transceiver Station BSC: Base Station Controller MSC & GMSC: Switching Centers HLR & VLR: Location Registers RSS: Radio Subsystem NSS: Network and switching subsystem OSS: Operation subsystem 22

23 NSS Databases Home Location Register (HLR) Contains permanent and semi-dynamic data of all subscribers Mobile subscriber ISDN number International mobile subscriber identity (IMSI) Current VLR and MSC Mobile subscriber roaming number (MSRN) This user-specific information exists once for each user in one HLR 23

24 NSS Databases Visitor Location Register (VLR) Local (dynamic) database for a subset of the user data Includes data about all users currently in the domain of the VLR When an MS enters into the area of the VLR, the VLR copies all relevant information from the HLR Avoid frequent HLR updates and long-distance signaling with the HLR VLR gets updated at MS startup and then periodically If no update is received by VLR for some time it's considered detached (e.g., roaming, new VLR takes over) 24

25 Question 6 How is localization, location update, roaming done in GSM and reflected in the databases? What are the typical roaming scenarios? (Optional) Looking at the HLR/VLR database approach used in GSM, how does this architecture limit the scalability in terms of users, and especially moving users? 25