Real-World Performance of current Mesh Protocols in a small-scale Dual-Radio Multi-Link Environment

Manuel Hachtkemper Michael Rademacher Karl Jonas [email protected] [email protected] [email protected]

22. ITG Fachtagung Mobilkommunikation May 10, 2017

1 Table of Contents

Introduction and motivation

Dual-radio mesh networks

Setups

Test procedure

Results

Conclusion

2 Introduction and motivation - Internet in rural areas

X Using a cost-efficent technology to bring connectivity to rural areas. I Local distribution of connectivity is the next step. I Dual-Radio WiFi Mesh Networks are (among others) one option: - Which mesh protocol to prefer? [Babel, B.A.T.M.A.N. V, BMX7, OLSRv2] - Which dual-radio setup to prefer?

Access: Internet WiFi Dual-Radio Mesh PSTN Core Network

Access Backhaul: WiFi Long-Distance Multi-Radio Mesh 3 Dual-radio mesh networks

1 Wireless

Wireless client

2 3 Channel A

Channel B

Ethernet

Internet 4 5

Example of a with two radios attached to each router.

4 Setup 1

Wireless router

Wireless client

Channel A

Ethernet

Traଏc source

Setup 1 for the experiments: One radio for everything on one channel; second radio unused. 5 Setup 2

Wireless router

Wireless client

Channel A

Channel B

Ethernet

Traଏc source

Setup 2 for the experiments: One radio for the mesh on one channel and another radio with a different channel for the clients. 6 Setup 3

Wireless router

Wireless client

Channel A

Channel B

Ethernet

Traଏc source

Setup 3 for the experiments: One channel for both mesh network and clients and a second radio with another channel for the mesh network. 7 Preliminary considerations for the experiments

I Has the system to “warm-up”? For how long?

I How to generate traffic? And for how long?

I How to get the measurement reproducible?

I How to prevent that different measurements affect each other?

8 Length of measurements

50000

45000

40000

35000

30000 data rate (kbit/s) data rate 25000

20000 0 100 200 300 400 500 600 time (s)

measurement 1 measurement 5 measurement 9 measurement 2 measurement 6 measurement 10 measurement 3 measurement 7 measurement 4 measurement 8

Development of the data rate over a period of 10 minutes. Intermediate values were taken every 10 seconds and always the overall data rate since the start is calculated. (Babel, Setup 3)

9 Length of measurements

Percentage deviation: x − x rel_dev = 20 10 ∗ 100 x10−x20 x 1.0 10 ● ● Deviation in reference to previous value 95% confidence interval ●

0.5 ● ● ●

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 0.0 ● ● ● ● ● ● ● ● Confidence interval: mean deviation (%) mean deviation −0.5 s [x ± t ∗ √ ] n−1,1−α/2 n −1.0

● 0 100 200 300 400 500 600 time (s) α = confidence level Development of the data rate over a period of 10 n = number of observations minutes. The mean percentage deviation in reference v u n to previous mean value is shown. (Babel, Setup 3) 1 X s = u (x − x)2 t n − 1 i i=1

10 Reproducibility / test procedure

C4 C5 1m

C3 1m 1m 1m C6 R2 R3

2m 1m 1m 1m Wireless router

C2 1m 1m Wireless client R1 R4

C1 1m

Tra¡ c source C1 Picture of the setup (in an underground Physical placement of nodes. parking lot).

11 Hypothesis

1. Using dual-radio routers compared to single-radio routers doubles the achievable data rate for clients.

I 2 channels = 2 * bandwidth = 2 * data rate

2. The mesh protocol influences the results, although all routers are direct neighbors.

I Different overhead for each protocol

3. Using both channels for the mesh (Setup 3) is worse than having a dedicated channel for all clients and one for the mesh (Setup 2).

I More mesh protocol overhead I The may use the channel which is occupied by the clients

12 Results: Single channel (S1) vs dual channel (S2)

40000

35000

30000

25000 data rate (kbit/s)

20000

15000 Babel S1 Babel S2 B.A.T.M.A.N.B.A.T.M.A.N. V S1 BMX7 V S2 S1 BMX7 S2 OLSRv2 S1OLSRv2 S2

Box plot of the results of Setup 1 and 2. Each box plot consists of ten measurements, where each data point is the sum of the six client results. 13 Dedicated access (S2) vs mixed mesh/access (S3)

40000

38000

36000

34000

32000

data rate (kbit/s) 30000

28000

26000 Babel S2 Babel S3 B.A.T.M.A.N.B.A.T.M.A.N. V S2 BMX7 V S3 S2 BMX7 S3 OLSRv2 S2OLSRv2 S3

Box plot of the results of Setup 2 and 3. Each box plot consists of ten measurements, where each data point is the sum of the six client results. 14 Conclusion

I Mesh protocols have specific features for multi-radio networks.

I Expected: Dual-radio routers = 2 * data rate of single-radio routers.

I Not expected: Different mesh protocols lead to similar results (in our scenario).

I Not expected: Using both radios within the mesh is equally good and should be preferred (in our scenario).

I The protocol overhead is negligible in small networks

15 Thank you very much!

Are there any questions?

Manuel Hachtkemper Michael Rademacher Karl Jonas [email protected] [email protected] [email protected] 16 References

[1] L. Cerdà-Alabern, A. Neumann, and L. Maccari. “Experimental Evaluation of BMX6 Routing Metrics in a 802.11an Wireless-Community Mesh Network”. In: Future Internet of Things and Cloud (FiCloud), 2015 3rd International Conference on. 2015, pp. 770–775. doi: 10.1109/FiCloud.2015.28.

[2] Open Mesh. Network Wide Multi Link Optimization (technical documentation). https://www.open-mesh.org/projects/batman-adv/wiki/Network-wide-multi-link-optimization. [Online; last visit 2016-11-20]. 2016.

[3] J. Chroboczek. Diversity Routing for the Babel Routing Protocol. Internet-Draft draft-chroboczek-babel-diversity-routing-00. IETF Secretariat, 2014. url: http://www.ietf.org/internet-drafts/draft-chroboczek-babel-diversity-routing-00.txt.

[4] Open Mesh. B.A.T.M.A.N. V. https://www.open-mesh.org/projects/batman-adv/wiki/BATMAN_V. [Online; last visit 2016-11-8]. 2016.

[5] G. Daneels. Analysis of the BMX6 Routing Protocol (Master’s Thesis). Belgium: University of Antwerp, 2013.

[6] J. Chroboczek. The Babel Routing Protocol. RFC 6126 (Experimental). Updated by RFCs 7298, 7557. Internet Engineering Task Force, Apr. 2011. url: http://www.ietf.org/rfc/rfc6126.txt.

[7] T. Clausen et al. The Optimized Link State Routing Protocol Version 2. RFC 7181 (Proposed Standard). Updated by RFCs 7183, 7187, 7188, 7466. Internet Engineering Task Force, Apr. 2014. url: http://www.ietf.org/rfc/rfc7181.txt.

[8] “IEEE Standard for Information Technology – Telecommunications and information exchange between systems – Local and metropolitan area networks – Specific requirements. Part 11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications”. In: IEEE Std 802.11-2012 (Revision of IEEE Std 802.11-2007) (2012), pp. 1–2793.

17