Energy Consumption in Ad-Hoc Routing Protocols: Comparing DSR, AODV, and Tora
| Yasser Gadallah and Thomas Kunz | |
| Carleton University | |
| http://kunz-pc.sce.carleton.ca/ | |
| tkunz@sce.carleton.ca |
| Mobile Devices constrained by many things, one of them being battery | ||
| Sending and receiving messages is one of the more power-expensive operations | ||
| Data packets | ||
| Control packets of routing protocol | ||
| Wireless Interfaces also consumer substantial amount of power while idle, but putting them into the low-power “sleep” mode has to be done carefully | ||
| Goal: study power consumption of popular MANET routing protocols | ||
| to understand their energy behaviour, and | ||
| to draw lessons for the design of more efficient routing protocols | ||
| What is an “energy-efficient” routing protocol? | ||
| Possible Metrics: | ||
| Minimize energy consumed/packet, | ||
| Maximize time to network partition, | ||
| Minimize variance in node energy levels, | ||
| Minimize cost/packet, and | ||
| Minimize maximum node cost. | ||
| We used: average overall amount of consumed energy | ||
| Note: if minimizing energy consumption was only goal, a trivial optimal solution exists: drop all packets J | ||
| Therefore, need to also consider packet delivery ratio | ||
| Assumption: only wireless interface consumes energy (i.e., internal processing of messages etc. is “free”) | |
| NS2 has (simplistic) energy model: “The energy model in a node has a initial value which is the level of energy the node has at the beginning of the simulation. This is known as initialEnergy. It also has a given energy usage for every packet it transmits and receives. These are called txPower and rxPower” | |
| Default IDLE energy consumption is zero, send/receive costs independent of packet size, unicast/broadcast, etc. |
| Re-implement energy model to correspond to published measurements of actual IEEE 802.11 interface: | ||
| Consumed energy per packet = m * (packet size) + b | ||
| m and b are empirical constants that are based on whether the packet is being sent, received, promiscuously handled or discarded. They also depend on the operation type (unicast vs. broadcast). | ||
| Idle power: 0.8 Watts | ||
Routing Protocols and Simulation Environment
| Routing Protocols: | ||
| AODV: on-demand protocol with traditional routing tables | ||
| DSR: source routing, route caches, nodes (optionally) listen promiscuously to learn information to update cache | ||
| TORA: link reversal algorithm, maintains enough info so that link failure not necessarily results in new route discovery, more complex control messages | ||
| Simulation Environment: “the usual” | ||
| NS2, 50 nodes, 1500 x 300 m, initial power: 1000 Joules | ||
| CBR sources, 5 packets/s, 512 byte packets, 12 senders | ||
| Random waypoint model, 20 m/s max speed, variable pause time | ||
| 600 seconds simulation length, 4 different movement patterns per scenario | ||
Analytical and Simulation Results
| Functional Comparison Results | ||
| protocol analysis: predict that AODV and DSR should perform similarly, with TORA being worse | ||
| Simulation Results: not quite that (at first sight) | ||
| Problem: TORA delivers far fewer packets (i.e., saves energy by not delivering them often enough at the destination!), based on the simulation conditions that we used |
Simulation Results: Energy Balance
Simulation Results:
Idle Energy Consumption Matters
| Conclusions: | ||
| AODV consumes slightly less energy than DSR at comparable packet delivery ratio | ||
| DSR seems to have slightly more balanced energy consumption between nodes (based on the difference (max - min) of nodal energy consumption) | ||
| Idle energy is about half of total energy consumed | ||
| Future Work: | ||
| Design protocol that is energy-efficient and fair (i.e., balanced) – work under way | ||
| Compare to other published energy efficient routing protocols such as SPAN, etc. | ||