Sie sind auf Seite 1von 3

COMP 8531 Final Exam

Harvey Le (A00811495) Instructor: Tejinder Randhawa Part 1 1. Enhanced Distributed Channel Access (EDCA) HCF Control Channel Access (HCCA) 2. HCCA is more suitable for real time applications because its an asynchronous real time scheduler 3. The 802.11b standard consists of a Distributed Coordination Function (DCF), which coordinated the usage of medium via the use of CSMA/SA Since multiple stations are connected and connections are full duplex, CSMA/CA should be employed in 802.11b 4. In 802.11b standard, SIFS is Short Inter Frame Spacing, which is the amount of time a station must wait before sending or starting to receive a RTS/ CTS or ACK, DIFS is Distributed IFS, which is the amount of time a station must wait before sense a clear radio before starting a new transmission sequence and PIFS is Point Coordination IFS, which is the DIFS for the access point in a special access method (Point Coordination Function) The reason for a packet to be transmitted after SIFS, DIFS or PIFS is that when the MAC receives a request for transmission, the medium is checked by the physical and virtual Carrier Sense mechanisms. If all of them indicate an idle medium, which is required, for a period of DIFS, SIFS or PIFS, then the transmissions are allowed to begin 5. RTS/CTS is employed because in a particular station, it will prevent that station from sending a data frame to another different station until the current handshake with the current station is completed Another reason for RTS/CTS to be employed is that it can minimize collision among hidden stations 6. The provisions exist in 802.11e for QoS support are DCF (Distributed Coordination Function) Part 2 1. In a Bluetooth piconet, a master is the one who controls the communication. Therefore, slaves are devices connected with the master 2. Up to 7 slaves are allowed for a master

3. A Bluetooth piconet can allow up to 7 simultaneously active slave devices or 255 slave devices in parked mode 4. A master of a piconet can only be a slave for another master in another piconet, cannot be a master. Part 3 1. The time for the node to wait is d = l/R where l is the distance between that node and node V, R is the Part 4 1. An example for necessitating fast indexing schemes for LBS involving moving objects is that most moving objects are constantly changing their position and submitting request to a database, which can reduce system performance. Thus, fast indexing mechanisms are needed 2. The main advantage of IMORS over TPR tree is that the update cost is about 40% less Part 5 1. TCP was initially designed and implemented for wired networks. Thus, in wireless networks, whose one characteristic is congestions or noise, TCP performance reduces. When congestions and transmission errors occur, the link-layer mechanisms will recover the losses, which are caused by transmission errors, and the transport protocol will recover the losses that are caused by congestions 2. In TCP-DCR, the response to the duplicate acknowledgements is delayed by a short period (T). If the packet is recovered by link-level retransmission before the end of T, TCP-DCR will proceed as if the packet loss never occurred. However, if the packet is not recovered by the end of T, TCP-DCR will start the congestion recovery algorithms of fast retransmission and recovery. Part 6 1. TCP Friendly Rate Control (TFRC): The receiver measures the loss event rate and send this information back to the sender. The the sender uses the feedback messages to measure the round-trip time (RTT). The loss event rate and RTT are fed into TFRCs throughput equation, giving the acceptable transmit rate. Finally, the sender adjusts its transmit rate to match the calculated rate. Multiple TFRC (MULTFRC): This approach is to measure the round-trip time (RTT) and adjust the number of connections accordingly to utilize the wireless bandwidth efficiently and ensure fairness between applications.

Part 7 1. An example of serializability: Lets consider we have 3 bank accounts (A, B, C). There are 2 transactions: first is to transfer 100 from A to B and second to to transfer 100 from A to C. A schedule that is considered serializable should be like this Transaction 1 will be done first Read (A, t) t = t - 100 Write (A, t) Read (B, t) t = t + 100 Write (B, t) Then transaction 2 starts Read (A, s) s = s - 100 Write (A, s) Read (C, s) s = s +100 Write (C, s) Those steps above must be done correctly, they are not allowed to overlap 2. The key issue relates to 2-tier replication, which allows user to run transactions locally and when user connects to server, those transactions will be re-executed at servers. The problem is user executes transactions without knowing other transactions are doing and this can lead to a huge number of transaction rejections

Das könnte Ihnen auch gefallen