Buddy Routing: A Routing Paradigm for NanoNets Based on Physical Layer Network Coding

Buddy Routing: A Routing Paradigm for NanoNets Based on

Physical Layer Network Coding

Ruiting Zhou+, Zongpeng Li+, Chuan Wu*, Carey Williamson+

+University of Calgary*University of Hong Kong

Outline

• 1. Introduction to NanoNets• 2. Enabling Buddy Routing • 3. Theoretical Analysis• 4. Buddy Routing: Unicast• 5. Buddy Routing: Multicast• 6. Conclusion

2／ 23

• 1. Introduction to NanoNets NanoNets Collaborative Data Forwarding

3／ 23

1.1 NanoNets • NanoNets:  Networks of

nanomachines at extremely small dimensions -- on the order of nanometers or micrometers.

• Basic computing and communication feasible on nanonodes.• Large network size and node density, low cost and available power.

4／ 23

1.2 Collaborative Data Forwarding

• Group nanonodes into collaborating pairs: overcome power constraint, enhance the communication range and rate.

• Enabled by physical layer network coding (PNC).

5／ 23

1.2 Collaborative Data Forwarding

• Enabled by Amplify&Forward (A&F).• Intra-pair transmission: transmit an

amplified version of the received analog signal:

6／ 23

• 2. Enabling Buddy Routing PNC vs A&F: Multi-hop

Transmission PNC vs A&F: One-hop BER

7／ 23

2.1 PNC vs A&F: Multi-hop Transmission

• PNC: Except at the source pair, no need for half-packet sharing.

• A&F: Intra-pair sharing of a half-packet is required at each hop, an extra step of transmission.

• lower end-to-end data throughput.

8／ 23

2.2 PNC vs A&F: One-hop BER• BER of PNC is almost the same as, but

slightly worse than, that of A&F, under the same SNR at the receiver side.

9／ 23

• 3. Theoretical Analysis System Model and Parameters Capacity and Power

Consumption

10／ 23

3.1 System Model and Parameters

• MAC layer protocol: TDMA• Two types of time slots:

Long time slot : long-hop data transmission happen simultaneously every three hops

Short time slot: all the intra-pair short hops transmit simultaneously

11／ 23

3.2 Capacity and Power Consumption

• We analyzed the end-to-end routing capacity of a BR route at very high SNR with and without noise considered.

• The extra power consumption overhead caused by BR is below 20%.

• The capacity is increased by a factor of 2.

12／ 23

• 4. Buddy Routing: Unicast BR Unicast Routing Algorithm Simulation

13／ 23

4.1 BR Unicast Algorithm

• Step 1: Pair-to-pair greedy geographic unicast routing. Forwarding process: it looks for a

next-hop pair between the two co-axial circles of radius and , which is closest to the destination.

If the last pair is too close to the destination, replaced by a new pair such that the distance between the new pair and the destination is just larger than .

and are the minimum and maximum allowed distances between two neighbor buddy pairs.

14／ 23

• Radius of red circle is the maximum distance between a pair of buddy nodes.

• When Pair 70 looks for the next hop towards the destination, it can only search the area in the blue ring.

4.1 BR Unicast Algorithm

15／ 23

• Step 2: Iterative MAC layer optimization. Adjust time slot: so that the

capacity in each time slot is equal. Inter-pair power optimization:

adjust the long hop transmission power — achieve equal capacity at bottleneck link & 2 neighbor links.

Intra-pair power optimization: adjust the short hop transmission power —achieve equal capacity at bottleneck pair & 2 neighbor pairs.

4.1 BR Unicast Algorithm

16／ 23

4.1 BR Unicast Algorithm

• End-to-end throughput is doubled after the iterative power/MAC optimization

17／ 23

4.2 Simulation• BR Unicast, end-to-end throughput comparison,

different x-axis.• Throughput of BR after optimization is almost

twice that of point-to-point routing

18／ 23

• 5. Buddy Routing: Multicast The Multicast BR Gadget BR Multicast Tree

19／ 23

• Multicast: has branches in the transmission topology.

• Replicate a data packet from an upstream node pair to two pairs.

5.1 The Multicast BR Gadget

20／ 23

5.2 BR Multicast Tree• One-to-four multicast

21／ 23

6. Conclusion• We proposed a new PNC-based

routing paradigm, Buddy Routing, for NanoNets.

• BR has a potential to break through the nodal power limit in NanoNets.

• It can substantially improve the unicast and multicast throughput, as verified by our theoretical analysis and simulation results 22／ 23

• Thanks!

• Questions?

23／ 23