Download presentation
Presentation is loading. Please wait.
Published byIrene Nichols Modified over 9 years ago
1
Daibo Liu 1 Daibo Liu 1, Zhichao Cao 2, Xiaopei Wu 2, Yuan He 2, Xiaoyu Ji 3 and Mengshu Hou 1 ICDCS, 2015, Columbus TeleAdjusting: Using Path Coding and Opportunistic Forwarding for Remote Control in WSNs University of Electronic Science and Technology of China 1 University of Electronic Science and Technology of China Tsinghua University 2 Tsinghua University Hong Kong University of Science and Technology 3 Hong Kong University of Science and Technology 1
2
Outline Motivation Related works and our approach Design of TeleAdjusting Implementation and Evaluation Summary of this work 2
3
Motivation 3 Practical experience – CitySee: city-wide urban sensing system – Predefined configurations VS. changes of network – Expensive manual maintenance Remote control in duty cycled WSNs – Key technique for network management – Challenge to achieve energy efficiency and reliability
4
Outline Motivation Related works and our approach Design of TeleAdjusting Implementation and Evaluation Summary of this work 4
5
Related Works Remote control protocols in WSNs – Unstructured approaches Network-wide flooding Energy and bandwidth waste, reliability guarantee Drip, Deluge, Gloosy – Structured approaches Along a predefined path Energy efficiency Susceptible to network dynamics, hard to guarantee reliability E.g., RPL, ORPL 5 Can we achieve a remote control approach guaranteeing both reliability and efficiency?
6
Forward through an Optimal Area The problems: – For each node, generating the optimal path from sink to it in distributed manner – Forward downwards along the predefined optimal path – Nodes around the optimal path help to forward – Guarantee both efficiency and reliability 6 Forwarding downwards means forwarding from sink to an individual node
7
Our weapons 7 Path coding – Encoded optimal path from sink to each node – Binary string implies the relationship between paths – Prefix-matching process for forwarding selection Opportunistic forwarding – Closer and earlier wake-up nodes help to forward
8
Outline Motivation Related works and our approach Design of TeleAdjusting Implementation and Evaluation Summary of this work 8
9
Design of TeleAdjusting Overview of design – Generating path code – Exploiting opportunistic forwarding 9 Upstream node denotes the next node in the path from it to the sink, and upstream nodes denote all nodes in the path from it to the sink.
10
Encoding Path Concept of path code – Reverse path address – 0-1 bit string with valid code length – Encoding certain relationship with other nodes – Parent’s code is the prefix of children nodes’ codes 10 One valid bit A and M are S’s children nodes, they set S’s valid codes as the prefix of their path codes. The same prefix The same prefix also indicates they are with the same parent.
11
Position allocation Encoding Path 11 Allocate a unique position to each children node Parent node’s path code Position space Path code = Prefix code + bit string position – Prefix code: valid path code of parent node – Position: uniquely allocated by parent node Generating path code
12
Encoding Path 12 Position allocation – Deterministic allocation – Against allocation data loss/new joining child node Position request and allocation acknowledgement – Position maintenance – Space extension Allocating a unique position to each children node Parent node’s path code Position space Generating path code
13
Opportunistic Forwarding Opportunistic forwarding in TeleAdjusting – Earlier wake-up & closer to destination will assist to forward – Metrics : prefix length (logical distance to destination) Information attached in control packet – Expected relay (E) and the valid path code length (L), the appointed destination (D) and its path code (π D ) Formalization – π: path code – F(π A, π B ): the identical prefix length of the path codes of A and B 13
14
Opportunistic Forwarding Forwarding condition: S will relay the overheard control packet IF any one of the three is satisfied – S == E, where E is the expected relay – F(π S, π D )> F(π E, π D ) – F(π N, π D )> F(π E, π D ), N is a neighbor of S Forwarding strategies – Along the predefined path – Exploiting available opportunities Earlier wake-up relay Closer to the destination – Backtrack if a node cannot forward downwards – Against unreachable problem 14
15
Forwarding Strategy 12 Along the predefined path – Without exploiting opportunistic forwarding – Traveling along the encoded path Encoded path Traveling path
16
Forwarding Strategy 16 Exploiting available opportunities Encoded path Traveling path – Relay in the encoded path but closer than the expected relay to the destination – Relay around the encoded path – At least one of the relay’s neighbors is closer to the destination and in the encoded path Efficiency: Opportunistic forwarding could exploit the earlier wake-up relays to forward control packet. Reliability: Exploiting opportunistic forwarding will increase the ability of resisting network dynamics.
17
Forwarding Strategy 17 Backtrack – If a node (E) can’t further forward towards the destination Operation: Set its potential relays (C and D) to unreachable Set itself to unreachable to the destination – If unreachable It will not actively forward the control packet – Reset to reachable if any potential relay is reachable Backtrack strategy will eventually find a path to the destination, otherwise, sink will set itself to unreachable.
18
Forwarding Strategy 18 Against unreachable problem – Sink node is set to unreachable – Report to the controller – Controller selects a neighbor (K) of the destination Maximum prefix difference – Sink forwards control packet to the appointed K – K forwards the control packet to the destination by unicast forwarding
19
19 TeleAdjusting in protocol stack Beneath application layer Above MAC layer Connecting with link estimator and network layer Integrate TeleAdjusting into protocol stack
20
Outline Motivation Related works and our approach Design of TeleAdjusting Implementation and Evaluation Summary of this work 20
21
Implementation and Evaluation 21 Implementation of TeleAdjusting – TinyOS-2.1.1 – Built upon LPL – Interface for application la yer Evaluation: – Large-scale simulation in TOSSIM – Indoor testbed, 40 TelosB nodes – Performance: reliability, efficiency – Comparison with Drip, quasi-RPL
22
Testbed Settings 22 40 TelosB sensor nodes – 22 nodes on the testbed board – 18 nodes scattered around the testbed Multi-hop networks 512ms wakeup interval Periodical remote control packet (10 minutes)
23
Simulation results 23 Path code length almost increases linearly with hop count, no matter tinght-grid or sparse- linear. Simulation setup – Sparse-linear: 5×45 grids with low gain – Tight-grid: 15x15 grids with high gain – Network topology construction: CTP+Trickle
24
Simulation results 24 Convergent time = Path code generated time – routing found time Reverse path hop count vs. CTP hop count Nodes can quickly generate its path code and associate different positions to children nodes almost without exceeding 20 beacons time. For each node, the reverse hop count is very close to its CTP routing hop count.
25
Evaluation Results 25 Remote packet delivery ratio – Two scenarios: interfered (WIFI) channel and clear channel – Comparison with Drip (reliability guarantee), quasi-RPL – Re-Tele is TeleAdjusting dealing with unreachable problem Structured approach (RPL) is susceptible to network dynamics, unstructured approach (Drip) guarantees the reliability of remote packet delivery. The reliability of TeleAdjusting is close to Drip.
26
Evaluation Results 26 Transmission hop count vs. CTP hop count Average network-wide transmission count
27
Outline Motivation Related works and our approach Design of TeleAdjusting Implementation and Evaluation Summary of this work 27
28
Summary Ready-to-use remote control protocol; Distributed coding based addressing scheme that encodes path information from sink to individual nodes; Exploiting opportunistic forwarding to guarantee both reliability and efficiency; Implementation of TeleAdjusting in TinyOS; Simulation and real testbed evaluation. 28
29
29 Thank you! Q&A Q&A
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.