Presentation is loading. Please wait.

Presentation is loading. Please wait.

Hongchao Zhou, Xiaohong Guan, Chengjie Wu

Similar presentations


Presentation on theme: "Hongchao Zhou, Xiaohong Guan, Chengjie Wu"— Presentation transcript:

1 Hongchao Zhou, Xiaohong Guan, Chengjie Wu
RTMC: Reliable Transport with Memory Consideration in Wireless Sensor Networks Hongchao Zhou, Xiaohong Guan, Chengjie Wu

2 Reliable Transport with Memory Consideration Experiments
Outline Background Reliable Transport with Memory Consideration Experiments Simulation and Comparison 2019/5/19

3 Wireless sensor networks
Wireless sensor networks (WSN) consist of many resource constrained sensor nodes with limited computational capability, communication radius and memory space. The availability of low-cost hardware such as CMOS cameras and microphones enables sensor nodes to capture images or audio from environment, and increases the requirement of reliable file transport. Image Audio Video 2019/5/19

4 Problems and Challenges
Requirement: the files to be transmitted generally divided into multiple segments to avoid high packet error rate. Problem: How to send all the multiple segments from a source to the sink via multiple unreliable hops? Challenges unreliable links in WSN and conflict in transmissions from different nodes may cause packet loss limited memory of relay sensor nodes with possible overflow may also cause packets loss. 2019/5/19

5 Existing Methods Node-by-node: One relay node receives all the segments and then re-transmit all the segments to the next hop. Possible large delivery latency Significant memory requirement TCP is widely used to provide end-to-end reliability and congestion control in Internet. End-to-end retransmission is not appropriate in wireless systems with high packet-loss rate. End-to-end congestion control may have a tardy response and cannot adapt to the rapid change of channel capacity in WSN. Control Messages may lost. 2019/5/19

6 A reliable transport protocol:
A reliable transport protocol in WSN should hop-by-hop retransmission hop-by-hop congestion control 100% reliability (some segments may very important) make good use of bandwidth Reliable Transport with Memory Consideration (RTMC) 2019/5/19

7 Reliable Transport with Memory Consideration Experiments
Outline Background Reliable Transport with Memory Consideration Experiments Simulation and Comparison 2019/5/19

8 RTMC Basic Ideas: If the local node knows that its next hop can store m extra segments, then it can send m segments without any negotiations. If the local node does not know whether one segment is received by the next hop, it can not remove this segment. Variable Definition segment_length : the number of segments to transport BL :maximal number of segments (small integer) local_free :extra segments can be stored in the local node next_free :extra segments can be stored in the next hop rec_segments : the IDs of the received last BL segments 2019/5/19

9 RTMC Packets: Initial Packet:
Ask the next hop to participate into the current transport. local_free, rec_segments, segment_length Data Packet: … Send a segment to the next hop local_free, rec_segments, segment_id, content Requiring Packet: Ask the last hop to send a data packet. local_free, rec_segments I 1 2 R 2019/5/19

10 RTMC An instance to demonstrate that how RTMC works. BL=3
segment_length=9 hops =3 2019/5/19

11 Example by RTMC Beginning Send initial packets 1 1 2 2 3 3 4 4 5 5 6 6
7 4 7 5 8 5 8 9 6 6 9 source sink A B I 2019/5/19

12 RTMC instance Beginning
If A does not receive any response from B, an initial message should be retransmitted. 1 1 2 2 3 3 4 4 5 5 6 6 4 4 7 7 5 8 5 8 9 6 6 9 source sink A B I I 2019/5/19

13 RTMC instance Transporting
Receive a data packet from the last hop node. local_free rec_segments 1 2 2 6-4-3 7-6-4 7-6-4 5 1 6 2 2 2 7 3 6 2 2 8 4 2 6 7 7 9 source sink A B 7 2019/5/19

14 RTMC instance Transporting Hear a data packet from the next hop node.
local_free rec_segments 1 2 7-6-4 7-6-4 5 next_free=2 1 6 ready to transmit 2 2 2 7 3 6 2 2 8 4 2 6 7 7 9 source sink A B 7 2019/5/19

15 RTMC instance Transporting
Special case: A is empty but source thinks that A is full, due to packets loss. A should send Requiring packets to source periodically until memory is not empty. next_free=0 next_free=2 1 3 7-6-4 7-6-4 5 1 6 ready to transmit 2 2 2 7 3 2 7 2 8 4 2 6 2 9 source sink A B R 2019/5/19

16 RTMC instance Ending For a node, if (1) its memory is empty and (2) it has received all of the segments, then the transport is finished of this node. 2019/5/19

17 Reliable Transport with Memory Consideration Experiments
Outline Background Reliable Transport with Memory Consideration Experiments Simulation and Comparison 2019/5/19

18 Experiment Scenario Mica2, 19.2kbps, 10K image file, 100segments
2019/5/19

19 Experiment result without congestion
Time (s) Segment sequence No. 2019/5/19

20 Experiment result with congestion
G was turned on 20seconds later 2019/5/19

21 Experiment result with congestion
Time (s) Segment sequence No. 2019/5/19

22 Reliable Transport with Memory Consideration Experiments
Outline Background Reliable Transport with Memory Consideration Experiments Simulation and Comparison 2019/5/19

23 Comparison SEA A receiver sends back an acknowledge after receiving a packet. The sender retransmits the packet if it does not receive the acknowledge. Each node can store no more than 10 segments RTMC Each node can store 5 segments Simulation 5 nodes, TDMA (one message for RTMC, one message and one acknowledge for SEA). p p p p source sink 2019/5/19

24 Simulation Result Transport time (periods) Packet success rate p
2019/5/19

25 Simulation Result Packets transmitted Packet success rate p 2019/5/19

26 Conclusions RTMC can get 100% reliability due to hop-by-hop retransmission and congestion control mechanism. RTMC can use channel effectively and has low delivery latency, since the transmission rate can quickly change with the traffic change in the network. RTMC costs less energy, since the number of retransmission packets is reduced. Both experiment and simulation results support these conclusions 2019/5/19

27 Thank you 2019/5/19


Download ppt "Hongchao Zhou, Xiaohong Guan, Chengjie Wu"

Similar presentations


Ads by Google