Energy Efficient and Enhanced Power Aware Routing for Mobile Ad Hoc Networks
11/07/001 Power-Aware Ad Hoc Routing Mobicom98 paper “Power-Aware Routing in Ad Hoc Networks” by...
-
date post
20-Dec-2015 -
Category
Documents
-
view
218 -
download
0
Transcript of 11/07/001 Power-Aware Ad Hoc Routing Mobicom98 paper “Power-Aware Routing in Ad Hoc Networks” by...
![Page 1: 11/07/001 Power-Aware Ad Hoc Routing Mobicom98 paper “Power-Aware Routing in Ad Hoc Networks” by Singh, Woo, and Raghavendra Presentation by – Harkirat.](https://reader035.fdocuments.us/reader035/viewer/2022062516/56649d415503460f94a1b41e/html5/thumbnails/1.jpg)
11/07/00 1
Power-Aware Ad Hoc Routing
Mobicom98 paper “Power-Aware Routing in Ad Hoc Networks” by Singh, Woo, and
Raghavendra
Presentation by – Harkirat Singh
![Page 2: 11/07/001 Power-Aware Ad Hoc Routing Mobicom98 paper “Power-Aware Routing in Ad Hoc Networks” by Singh, Woo, and Raghavendra Presentation by – Harkirat.](https://reader035.fdocuments.us/reader035/viewer/2022062516/56649d415503460f94a1b41e/html5/thumbnails/2.jpg)
11/07/00 2
Motivation• Conservation of power and power-aware routing must take
into consideration • Low Power Hardware - energy efficient displays - CPU’s with active and doze mode - low-power I/O devices• Energy-efficient algorithms• 40-70% saving in energy using PAMAS• Other factors - average battery life in idle cellular phone – one day - DEC Roamabout 5.76 W transmission, 2.88 W receive,
0.35 W idle
![Page 3: 11/07/001 Power-Aware Ad Hoc Routing Mobicom98 paper “Power-Aware Routing in Ad Hoc Networks” by Singh, Woo, and Raghavendra Presentation by – Harkirat.](https://reader035.fdocuments.us/reader035/viewer/2022062516/56649d415503460f94a1b41e/html5/thumbnails/3.jpg)
11/07/00 3
Listening Problem
B A C A transmits A’s transmission is to B overheard by C
![Page 4: 11/07/001 Power-Aware Ad Hoc Routing Mobicom98 paper “Power-Aware Routing in Ad Hoc Networks” by Singh, Woo, and Raghavendra Presentation by – Harkirat.](https://reader035.fdocuments.us/reader035/viewer/2022062516/56649d415503460f94a1b41e/html5/thumbnails/4.jpg)
11/07/00 4
The Problem• Shortest-hop rotes (green, blue, red) all use middle (black) node’s resources. It’s battery will die early.• Fairness issue• Routing packets through lightly loaded nodes also helps in energy expended in contention.
![Page 5: 11/07/001 Power-Aware Ad Hoc Routing Mobicom98 paper “Power-Aware Routing in Ad Hoc Networks” by Singh, Woo, and Raghavendra Presentation by – Harkirat.](https://reader035.fdocuments.us/reader035/viewer/2022062516/56649d415503460f94a1b41e/html5/thumbnails/5.jpg)
11/07/00 5
Goal:
Reduce the energy consumption of whole communication system
Increase lifetime of nodes/network until
partition
![Page 6: 11/07/001 Power-Aware Ad Hoc Routing Mobicom98 paper “Power-Aware Routing in Ad Hoc Networks” by Singh, Woo, and Raghavendra Presentation by – Harkirat.](https://reader035.fdocuments.us/reader035/viewer/2022062516/56649d415503460f94a1b41e/html5/thumbnails/6.jpg)
11/07/00 6
Brief overview of Routing Protocols
• On-Demand Routing protocols
- no up-to-date routes are maintained
- Routes are created as an when basis:
call passive convergence.• Table-Driven Routing protocols
- Each node maintains a table containing routing
information to every other node in network:
call active convergence.
![Page 7: 11/07/001 Power-Aware Ad Hoc Routing Mobicom98 paper “Power-Aware Routing in Ad Hoc Networks” by Singh, Woo, and Raghavendra Presentation by – Harkirat.](https://reader035.fdocuments.us/reader035/viewer/2022062516/56649d415503460f94a1b41e/html5/thumbnails/7.jpg)
11/07/00 7
Dynamic source Routing
• Source-routed on-demand routing protocol
1
6
8
7
5
3
2
4
Src
1
6
8
7
5
3
2
4
Src
<1><1,2>
<1>
<1,3>
<1,3,5>
<1,3,5,7>
<1><1,4>
<1,4,6>
Dest.Dest
<1,4,6>
<1,4,6>
<1,4,6>
Route Discover – Route Record Propogation of Route Reply with Route Record
![Page 8: 11/07/001 Power-Aware Ad Hoc Routing Mobicom98 paper “Power-Aware Routing in Ad Hoc Networks” by Singh, Woo, and Raghavendra Presentation by – Harkirat.](https://reader035.fdocuments.us/reader035/viewer/2022062516/56649d415503460f94a1b41e/html5/thumbnails/8.jpg)
11/07/00 8
Temporally-Ordered Routing Algorithm (TORA)
• Assign each node a “height” heuristic value based on various attributes
• A node with higher height called upstream and a node with lower height called downstream
• Route creation is done using QRY and UPD packets, results in DAG.
• In case of node failure, TORA floods a CLR (clear) packet.
![Page 9: 11/07/001 Power-Aware Ad Hoc Routing Mobicom98 paper “Power-Aware Routing in Ad Hoc Networks” by Singh, Woo, and Raghavendra Presentation by – Harkirat.](https://reader035.fdocuments.us/reader035/viewer/2022062516/56649d415503460f94a1b41e/html5/thumbnails/9.jpg)
11/07/00 9
TORA – Route creation• Height of destination assigned to 0 and all other
set to NULL(i.e. undefined)• Src broadcast a QRY packet with Dest id in it.
1
6
8
7
5
3
2
4
Src.
(-,-) (-,-)
(0,0)(-,-)
(-,-)
(-,-)
(-,-)
(-,-)
1
6
8
7
5
3
2
4
.
(0,3) (0,1)
(0,0)(0,2)
(0,1)(0,2)
(0,3)
(-,3)
Propogation of QRY packet UPD pkt – height updates
![Page 10: 11/07/001 Power-Aware Ad Hoc Routing Mobicom98 paper “Power-Aware Routing in Ad Hoc Networks” by Singh, Woo, and Raghavendra Presentation by – Harkirat.](https://reader035.fdocuments.us/reader035/viewer/2022062516/56649d415503460f94a1b41e/html5/thumbnails/10.jpg)
11/07/00 10
Spine routing Algorithm (SRA)
• Nodes are assigned to cluster
• Clusters are joined together by a virtual backbone.
• Reduce the complexity of maintaining routes
![Page 11: 11/07/001 Power-Aware Ad Hoc Routing Mobicom98 paper “Power-Aware Routing in Ad Hoc Networks” by Singh, Woo, and Raghavendra Presentation by – Harkirat.](https://reader035.fdocuments.us/reader035/viewer/2022062516/56649d415503460f94a1b41e/html5/thumbnails/11.jpg)
11/07/00 11
Ad Hoc Routing Protocolsand Usual Metrics
.
![Page 12: 11/07/001 Power-Aware Ad Hoc Routing Mobicom98 paper “Power-Aware Routing in Ad Hoc Networks” by Singh, Woo, and Raghavendra Presentation by – Harkirat.](https://reader035.fdocuments.us/reader035/viewer/2022062516/56649d415503460f94a1b41e/html5/thumbnails/12.jpg)
11/07/00 12
Contribution ??
• New Metrics• Minimize energy consumed per packet• Maximize time to network partition• Minimize variance in node power levels• Minimize cost per packet• Minimize maximum Node cost
![Page 13: 11/07/001 Power-Aware Ad Hoc Routing Mobicom98 paper “Power-Aware Routing in Ad Hoc Networks” by Singh, Woo, and Raghavendra Presentation by – Harkirat.](https://reader035.fdocuments.us/reader035/viewer/2022062516/56649d415503460f94a1b41e/html5/thumbnails/13.jpg)
11/07/00 13
Minimize Energy consumed/packet • Energy consumed for a packet j during traversal over
nodes n1.. nk
ej = Σ T(ni, ni+1 )
where T(a,b) denote the energy consumed per packet over one hop from a to b
Goal : Minimize ej packets j
Light loads same as shortest hop (assumes only variation in energy per hop is due to contention)
Tend to routes around congested areas (inc. hop count). Drawback : nodes with widely differing energy
consumption
K-1
i = 1
![Page 14: 11/07/001 Power-Aware Ad Hoc Routing Mobicom98 paper “Power-Aware Routing in Ad Hoc Networks” by Singh, Woo, and Raghavendra Presentation by – Harkirat.](https://reader035.fdocuments.us/reader035/viewer/2022062516/56649d415503460f94a1b41e/html5/thumbnails/14.jpg)
11/07/00 14
Maximize Time to Network Partition• Maximum-flow-min-cut
theorem gives a minimal set of nodes (the cut-set) the removal of which will cause network partition.
• Load balancing among cut-set nodes to ensure equal power drain.
S T
• Since nodes in different partitions independently determine routes, it’s hard to achieve global balancing while keeping low delay• Can not decide optimal path without knowledge of packet
![Page 15: 11/07/001 Power-Aware Ad Hoc Routing Mobicom98 paper “Power-Aware Routing in Ad Hoc Networks” by Singh, Woo, and Raghavendra Presentation by – Harkirat.](https://reader035.fdocuments.us/reader035/viewer/2022062516/56649d415503460f94a1b41e/html5/thumbnails/15.jpg)
11/07/00 15
Minimize variance in node power levels
• All nodes are equally important• Join Shortest Queue (JSQ) • RR if packets of same length
Minimize Cost / Packet• Goal : Maximize life of all the nodes in the
network• Nodes with depleted energy reserve should not lie
on many paths
![Page 16: 11/07/001 Power-Aware Ad Hoc Routing Mobicom98 paper “Power-Aware Routing in Ad Hoc Networks” by Singh, Woo, and Raghavendra Presentation by – Harkirat.](https://reader035.fdocuments.us/reader035/viewer/2022062516/56649d415503460f94a1b41e/html5/thumbnails/16.jpg)
11/07/00 16
Minimize Cost/Packet cont.
cj = Σ fi (xi ), where
fi (xi ) denotes the node cost of node i (node’s reluctance
to forward packet)
xi could be the energy consumed by node so far
cj represents the cost of sending packet j from node n1to nk
via intermediate nodes n2 .. nk-1
fi could reflect battery life remaining
Goal : Minimize cj, packets j
k-1
i = 1
![Page 17: 11/07/001 Power-Aware Ad Hoc Routing Mobicom98 paper “Power-Aware Routing in Ad Hoc Networks” by Singh, Woo, and Raghavendra Presentation by – Harkirat.](https://reader035.fdocuments.us/reader035/viewer/2022062516/56649d415503460f94a1b41e/html5/thumbnails/17.jpg)
11/07/00 17
Cost Function
fi(zi) = 1
(Zi – 2.8)
Where zi denotes measured voltage
fi ensures that shortest-hop routingWill be used when network is new
But as node approaches near end Of the lifetime, carefully routepackets to ensure longevity ofthese nodes
Lithium ion discharge curve
![Page 18: 11/07/001 Power-Aware Ad Hoc Routing Mobicom98 paper “Power-Aware Routing in Ad Hoc Networks” by Singh, Woo, and Raghavendra Presentation by – Harkirat.](https://reader035.fdocuments.us/reader035/viewer/2022062516/56649d415503460f94a1b41e/html5/thumbnails/18.jpg)
11/07/00 18
Benefits• It is possible to incorporate the battery characteristics
directly into the routing protocol
• Reduce variation in node cost and increase time to network partition
• Effects of network congestion are incorporated
Minimize Maximum Node Cost• Goal is to minimize maximum node cost after routing N
packets to their destinations or after T seconds
![Page 19: 11/07/001 Power-Aware Ad Hoc Routing Mobicom98 paper “Power-Aware Routing in Ad Hoc Networks” by Singh, Woo, and Raghavendra Presentation by – Harkirat.](https://reader035.fdocuments.us/reader035/viewer/2022062516/56649d415503460f94a1b41e/html5/thumbnails/19.jpg)
11/07/00 19
Overview of PAMASThis work assumes a MAC layer solution in which nodes
power off when can not transmit• Assumes separate signalling channel for RTS/CTS
exchange• RTS/CTS contain info about length of packet• Other nodes in neighborhood can predict how long to turn
off (no power wasted in listening)• Delays and throughput remain unchanged• Related work : (pagers) base transmits beacons and
minislot with ID of nodes with message waiting Other turn off. Reservation in 802.11(scheduling and
reservation better than contention).
![Page 20: 11/07/001 Power-Aware Ad Hoc Routing Mobicom98 paper “Power-Aware Routing in Ad Hoc Networks” by Singh, Woo, and Raghavendra Presentation by – Harkirat.](https://reader035.fdocuments.us/reader035/viewer/2022062516/56649d415503460f94a1b41e/html5/thumbnails/20.jpg)
11/07/00 20
Simulation and results
• Compared the performance of shortest-hop routing with shortest-cost routing in terms of :
- End-to-end packet delay
- Average cost/packet
- Average maximum node cost (after 300 sec)
• Used 16-node mesh topology and 10 and 20 node random graph
• Each simulation ran for 20 times, computed the mean and standard deviation for each of the three aforesaid metrics
![Page 21: 11/07/001 Power-Aware Ad Hoc Routing Mobicom98 paper “Power-Aware Routing in Ad Hoc Networks” by Singh, Woo, and Raghavendra Presentation by – Harkirat.](https://reader035.fdocuments.us/reader035/viewer/2022062516/56649d415503460f94a1b41e/html5/thumbnails/21.jpg)
11/07/00 21
% reduction in average cost in random network
20-node random network 10-node random network • Savings are greater in larger networks because larger networks
has more routes to choose from• Saving increases with load as cost differential increases, however, at very heavy loads it becomes constant (contention)
![Page 22: 11/07/001 Power-Aware Ad Hoc Routing Mobicom98 paper “Power-Aware Routing in Ad Hoc Networks” by Singh, Woo, and Raghavendra Presentation by – Harkirat.](https://reader035.fdocuments.us/reader035/viewer/2022062516/56649d415503460f94a1b41e/html5/thumbnails/22.jpg)
11/07/00 22
% reduction in maximum node cost
20-node random network 10-node random network
• Saving is more in denser network and increases with
![Page 23: 11/07/001 Power-Aware Ad Hoc Routing Mobicom98 paper “Power-Aware Routing in Ad Hoc Networks” by Singh, Woo, and Raghavendra Presentation by – Harkirat.](https://reader035.fdocuments.us/reader035/viewer/2022062516/56649d415503460f94a1b41e/html5/thumbnails/23.jpg)
11/07/00 23
% reduction in cost/pkt/hop and max node cost
% improvement in cost/pkt/hop % improvement in max. Node Cost
Saving in cost/packet increases with load and then decreases Because as load increases there is increasing difference in node costbetween two algos, later on costs are same so no saving
![Page 24: 11/07/001 Power-Aware Ad Hoc Routing Mobicom98 paper “Power-Aware Routing in Ad Hoc Networks” by Singh, Woo, and Raghavendra Presentation by – Harkirat.](https://reader035.fdocuments.us/reader035/viewer/2022062516/56649d415503460f94a1b41e/html5/thumbnails/24.jpg)
11/07/00 24
Conclusion
• Larger networks have higher cost saving,• Cost savings are best at moderate network
loads and negligible at low & very high loads,
• Denser networks exhibits more cost saving• It is easy to incorporate these metrics in
existing routing protocols for Ad Hoc Networks