27
“Controlled Straight Mobility and Energy-Aware Routing in Robotic Wireless Sensor NetworksRafael Falcon, Hai Liu, Amiya Nayak and Ivan Stojmenovic www.site.uottawa.ca/~ivan

“Controlled Straight Mobility and Energy-Aware Routing in Robotic Wireless Sensor Networks ” Rafael Falcon, Hai Liu, Amiya Nayak and Ivan Stojmenovic ivan

Embed Size (px)

Citation preview

Page 1: “Controlled Straight Mobility and Energy-Aware Routing in Robotic Wireless Sensor Networks ” Rafael Falcon, Hai Liu, Amiya Nayak and Ivan Stojmenovic ivan

“Controlled Straight Mobility and Energy-Aware Routing

in Robotic Wireless Sensor Networks”Rafael Falcon, Hai Liu,

Amiya Nayak and Ivan Stojmenovic

www.site.uottawa.ca/~ivan

Page 2: “Controlled Straight Mobility and Energy-Aware Routing in Robotic Wireless Sensor Networks ” Rafael Falcon, Hai Liu, Amiya Nayak and Ivan Stojmenovic ivan

Presentation Outline• Motivation and Problem Statement• Existing Solutions• Our Approach• Dynamic Optimal Progress Routing• Depth First Search (DFS)-based Routing• Routing for Disconnected Endpoints• Move Directly (MD)• Experiments• Conclusions and Future Work

Page 3: “Controlled Straight Mobility and Energy-Aware Routing in Robotic Wireless Sensor Networks ” Rafael Falcon, Hai Liu, Amiya Nayak and Ivan Stojmenovic ivan

Problem Statement

Assumptions:• Fixed source (S) and destination (D)• Long-term traffic (e.g. video surveillance)• Mobile relays (sensors, robots, vehicles)Problem: • Find a route from S to D.• Move each relay node to an optimal spot.• Minimize TX power and moving distance.

Page 4: “Controlled Straight Mobility and Energy-Aware Routing in Robotic Wireless Sensor Networks ” Rafael Falcon, Hai Liu, Amiya Nayak and Ivan Stojmenovic ivan

Prolonging Network Lifetime

Two commonly used methodologies:

1. Controlled Node Mobility

2. Energy-Aware Routing

Page 5: “Controlled Straight Mobility and Energy-Aware Routing in Robotic Wireless Sensor Networks ” Rafael Falcon, Hai Liu, Amiya Nayak and Ivan Stojmenovic ivan

Prolonging Network Lifetime• Our proposal:

– A hybrid routing-mobility frameworkfor the optimization of network communications.

1.Find energy-aware route from S to D.2.Move relay nodes from current to optimal positions in

straight line while preserving their connectivity along the way.

Page 6: “Controlled Straight Mobility and Energy-Aware Routing in Robotic Wireless Sensor Networks ” Rafael Falcon, Hai Liu, Amiya Nayak and Ivan Stojmenovic ivan

More Assumptions• Common communication radius r.• TX energy cost model is where d is distance.• Each node knows its own location.• Each node learns location of its 1-hop neighbors via

periodical “HELLO” messages.• Mobility cost is proportional to moving distance.

Page 7: “Controlled Straight Mobility and Energy-Aware Routing in Robotic Wireless Sensor Networks ” Rafael Falcon, Hai Liu, Amiya Nayak and Ivan Stojmenovic ivan

Existing Solutions

Goldenberg, Lin, Morse, Rosen and Yang: “Towards mobility as a network control primitive”, MOBIHOC, pp. 163-174, 2004.

• First controlled mobility approach.• Finds a route to destination D.• Relay nodes move in rounds (MR) to their optimal spots along the S-D straight line for energy saving.

[GLMRY04]

Page 8: “Controlled Straight Mobility and Energy-Aware Routing in Robotic Wireless Sensor Networks ” Rafael Falcon, Hai Liu, Amiya Nayak and Ivan Stojmenovic ivan

Existing SolutionsThe two routing algorithms used in [GLMRY04]

Greedy (forward to neighbor

closest to destination)

NP (forward to nearest

neighbor with progress to D)

Page 9: “Controlled Straight Mobility and Energy-Aware Routing in Robotic Wireless Sensor Networks ” Rafael Falcon, Hai Liu, Amiya Nayak and Ivan Stojmenovic ivan

Existing SolutionsMove in Rounds (MR)

S

D

• Each hop iteratively moves to the midpoint of its upstream and downstream neighbors.• Repeat until convergence.

Illustration of one movement round:

Page 10: “Controlled Straight Mobility and Energy-Aware Routing in Robotic Wireless Sensor Networks ” Rafael Falcon, Hai Liu, Amiya Nayak and Ivan Stojmenovic ivan

Existing Solutions

• Initial route is not energy-efficient.• Greedy and NP may fail in sparse networks.• Iterative node movement in rounds requires multiple

synchronization messages and causes unnecessary zigzag movement.• Large delay and possible communication failures.

Problems with [GLMRY04]:

Page 11: “Controlled Straight Mobility and Energy-Aware Routing in Robotic Wireless Sensor Networks ” Rafael Falcon, Hai Liu, Amiya Nayak and Ivan Stojmenovic ivan

Existing Solutions

Chen, Jiang and Wu: “Mobility control schemes with quick convergence in WSNs”, IPDPS, pp. 1-7, 2008.

• Improve MR with two more advanced mobility schemes (still in rounds)• Straight mobility never considered as it can break path connectivity during node movement.

[CJW08]

Page 12: “Controlled Straight Mobility and Energy-Aware Routing in Robotic Wireless Sensor Networks ” Rafael Falcon, Hai Liu, Amiya Nayak and Ivan Stojmenovic ivan

Our Approach1. Initial route is built in an energy-efficient way.

• Routing algorithm ensures message delivery.•Depth First Search (DFS) routing.

• Even target scenarios where S and D are disconnected!!• By collecting k relay nodes and dispatching them to their final locations. •k is optimal hop count.

Page 13: “Controlled Straight Mobility and Energy-Aware Routing in Robotic Wireless Sensor Networks ” Rafael Falcon, Hai Liu, Amiya Nayak and Ivan Stojmenovic ivan

Our Approach2. Relay nodes move directly (in straight line) to their

optimal locations while preserving the path connectivity as they go.

• Path connectivity can be maintained if the relay nodes coordinate among themselves prior to movement.

• Two connectivity preservation approaches are presented.

Page 14: “Controlled Straight Mobility and Energy-Aware Routing in Robotic Wireless Sensor Networks ” Rafael Falcon, Hai Liu, Amiya Nayak and Ivan Stojmenovic ivan

Our Approach

Page 15: “Controlled Straight Mobility and Energy-Aware Routing in Robotic Wireless Sensor Networks ” Rafael Falcon, Hai Liu, Amiya Nayak and Ivan Stojmenovic ivan

Power-Aware Routing• Many power-aware routing algorithms in the literature.

Liu, Nayak and Stojmenovic: “Localized mobility control routing in robotic sensor wireless networks”, MSN, pp. 19-31, 2007.[LNS07]

• Minimum Power-over-Progress Routing (MPoPR):• Optimal Hop Count Routing (OHCR):• round to k• select neighbour with distance closest to d(s,t)/k•More than optimal number of nodes selected

Page 16: “Controlled Straight Mobility and Energy-Aware Routing in Robotic Wireless Sensor Networks ” Rafael Falcon, Hai Liu, Amiya Nayak and Ivan Stojmenovic ivan

Dynamic Optimal Progress Routing (DOPR)•compute the optimal hop count k •Let U be current node and p the current hop count.• Compute dynamic optimal progress as:Select as next node the neighbor V such that:

• DOPR fails is no such neighbor is found.

Page 17: “Controlled Straight Mobility and Energy-Aware Routing in Robotic Wireless Sensor Networks ” Rafael Falcon, Hai Liu, Amiya Nayak and Ivan Stojmenovic ivan

Depth First Search (DFS) Routing

•Memorization is required at each node.• How does it work?• Each node sorts its neighbors according to

a particular selection criterion.• Packet is sent to top node in the list.• A visited node always rejects the packet• The sender then tries the next node in the list.

• DFS fails only when S and D are not connected.

Page 18: “Controlled Straight Mobility and Energy-Aware Routing in Robotic Wireless Sensor Networks ” Rafael Falcon, Hai Liu, Amiya Nayak and Ivan Stojmenovic ivan

Depth First Search (DFS) Routing

• We embed DOPR’s selection criterion into the DFS routing machinery.• The resulting power-aware algorithm is DOPR-DFS.• It behaves exactly as DOPR if no greedy failure occurs.• Otherwise, it keeps sending the packet to the remaining neighbors in the local sorted list of the current node.• Other hybrids: MPoPR-DFS, OHCR-DFS.

Page 19: “Controlled Straight Mobility and Energy-Aware Routing in Robotic Wireless Sensor Networks ” Rafael Falcon, Hai Liu, Amiya Nayak and Ivan Stojmenovic ivan

Routing for Disconnected Endpoints

• CKNR: Collect k neighbors via DFS routing.• Send them to their final locations along S – D line.• DOPR-CKNR, OHCR-CKNR, MPoPR-CKNR, etc.

S

D

Page 20: “Controlled Straight Mobility and Energy-Aware Routing in Robotic Wireless Sensor Networks ” Rafael Falcon, Hai Liu, Amiya Nayak and Ivan Stojmenovic ivan

Move Directly (MD)

• Moves the relay nodes straight (concurrently, in just one round) to their final energy-saving locations.• MD yields highest profit in spared TX power.• MD yields fastest convergence rate.• But… path connectivity can be broken during concurrent relay node movement.• Yet… it can be maintained if we introduce some degree of inter-relay coordination.• Two different approaches (proofs in the paper).

Page 21: “Controlled Straight Mobility and Energy-Aware Routing in Robotic Wireless Sensor Networks ” Rafael Falcon, Hai Liu, Amiya Nayak and Ivan Stojmenovic ivan

Distance-Free Path Preservation

• Hops are initially and finally connected (among themselves and with S and D).• Hops agree on same departure and arrival times.• Hops move at constant, individual speeds.

S

D

Page 22: “Controlled Straight Mobility and Energy-Aware Routing in Robotic Wireless Sensor Networks ” Rafael Falcon, Hai Liu, Amiya Nayak and Ivan Stojmenovic ivan

Distance-Bound Path Preservation

• Hops are initially and finally connected (among themselves and with S and D).• Initial and final internodal separation <=• Hops agree on same departure time.• Hops move at constant speed.

S

D

2/2R2/2R2/2R2/2R

2/2R

2/2R2/2R

2/2R

Page 23: “Controlled Straight Mobility and Energy-Aware Routing in Robotic Wireless Sensor Networks ” Rafael Falcon, Hai Liu, Amiya Nayak and Ivan Stojmenovic ivan

Experiments• DOPR is more energy-efficient than OHCR, MPoPR, Greedy and NP.• It also yields superior mobility vs. TX power gains.

Page 24: “Controlled Straight Mobility and Energy-Aware Routing in Robotic Wireless Sensor Networks ” Rafael Falcon, Hai Liu, Amiya Nayak and Ivan Stojmenovic ivan

Experiments• Empirical results confirm the feasibility of the proposed path connectivity preservation schemes.• Routing algorithms with larger inter-nodal distances

(e.g. Greedy) are more likely to break links among relay nodes during their advance to final locations.

Page 25: “Controlled Straight Mobility and Energy-Aware Routing in Robotic Wireless Sensor Networks ” Rafael Falcon, Hai Liu, Amiya Nayak and Ivan Stojmenovic ivan

Conclusions• A unified routing-mobility framework for the optimization of network communications has been proposed.• DOPR yields increased energy savings among peer routing protocols.• Message delivery is guaranteed in both sparse and dense network topologies via power-aware DFS.• Communication in partitioned networks made possible with CKNR.•Path connectivity was preserved with MD under two different scenarios.

Page 26: “Controlled Straight Mobility and Energy-Aware Routing in Robotic Wireless Sensor Networks ” Rafael Falcon, Hai Liu, Amiya Nayak and Ivan Stojmenovic ivan

Future Work

• To target more complex environments (e.g. multiple independent routing flows, multicasting, etc.)

• To quantify the coordination overhead required by MD.

Page 27: “Controlled Straight Mobility and Energy-Aware Routing in Robotic Wireless Sensor Networks ” Rafael Falcon, Hai Liu, Amiya Nayak and Ivan Stojmenovic ivan

Wireless Sensor Networks