Simple Modifications in HWMP for Wireless Mesh Networks with Smart Antennas Muhammad Irfan Rafique, Marco Porsch, Thomas Bauschert Chair for Communication Networks, TU Chemnitz irfan.rafique@etit.tu-chemnitz.de Chair for Communication Networks, TU Chemnitz EUNICE 2011 Page 1
Contents Introduction Motivation of the Work MAC Layer Amendments Path Selection (Routing) Algorithm Modified HWMP (MHWMP) Results Conclusion Chair for Communication Networks, TU Chemnitz EUNICE 2011 Page 2
Introduction - Wireless Mesh Networks [1] Chair for Communication Networks, TU Chemnitz EUNICE 2011 Page 3
Introduction - Hybrid Wireless Mesh Protocol (HWMP) Introduced in the IEEE 802.11s Draft Performs path selection (routing) at layer 2 Combines reactive and proactive part Uses a radio-condition aware routing metric (Airtime Metric ~ expected transmission time): C = O ca + O p + B r t 1 1 e fr O ca B t and are channel access and MAC protocol overheads is the number of bits of a test frame r O p is the transmission rate e fr denotes the frame error probability Chair for Communication Networks, TU Chemnitz EUNICE 2011 Page 4
Introduction - Performance Limitations of WMNs Performance problems in WMS arise due to: Multi-hop transmission Fluctuating wireless channel quality Possible improvement: Use of smart antennas instead of unidirectional antennas to improve transmission performance smart antennas consist of multiple antenna elements - they phase-shift the transmitted/received signal and scale it by weights Chair for Communication Networks, TU Chemnitz EUNICE 2011 Page 5
Introduction - Smart Antenna Transmission Schemes Spatial Multiplexing transmits multiple parallel signals increases data rate useful in dense mesh networks [2] Spatial Diversity transmits/receives the same signal from multiple antennas increases reliability higher transmission range appropriate in sparse mesh networks Beamforming forms a pattern of constructive and destructive interference increases reliability higher transmission range cancels interference parallel communication possible advantageous in sparse networks Chair for Communication Networks, TU Chemnitz EUNICE 2011 Page 6
Motivation of the Work State of the art in using smart antennas for WMNs: Current proposals employ only some transmission features of smart antennas do not fully exploit the potential of smart antenna to cope with dynamic channel conditions Target of our work: design of an cross-layer aware path selection algorithm (modified HWMP = MHWMP) that fully incorporates the advantages of smart antenna transmission schemes to adapt to dynamic channel conditions Key features of our algorithm (MHWMP): transmission of control frames with spatial diversity (STBC [4]) transmission of data frames with spatial multiplexing (MUX [5]) or beamforming (BF) depending on the channel conditions directional network allocation vector (DNAV) two neighboring and path tables Chair for Communication Networks, TU Chemnitz EUNICE 2011 Page 7
MAC Layer Amendments Challenge: cope with the range asymmetry problem of different transmission schemes STBC range omnidirectional antenna range In MHWMP: standard RTS/CTS prior to transmission with MUX RTS/CTS with STBC for beamforming transmission Chair for Communication Networks, TU Chemnitz EUNICE 2011 Page 8
MAC Layer Amendments Challenge: cope with the directional interference characteristic of beamforming In MHWMP: Directional Network Allocation Vector (DNAV) [6] to enable parallel communication feature of beamforming Chair for Communication Networks, TU Chemnitz EUNICE 2011 Page 9
Routing Algorithm Neighbor Discovery Challenge: neighborship depend on the transmission scheme In MHWMP: each node has two Neighboring Tables (NTs) Omnidirectional NT (ONT) neighbor discovery through normal beacons and handshakes used for MUX Directional NT (DNT) neighbor discovery with long range beacons and handshakes (with STBC) neighbors are stored together with their direction (position) used for BF Chair for Communication Networks, TU Chemnitz EUNICE 2011 Page 10
Routing Algorithm Path Discovery Two different path discoveries are used for the same destination each node has two Path Tables (PTs) Omnidirectional PT (OPT) normal PREQ/PREP with omnidirectional antenna (O-PREQ/O-PREP) max. MUX transmission rate r mux = min( M, N) r used for MUX Directional PT (DPT) PREQ with STBC (S-PREQ) Directional PREP (D-PREP) max. BF transmission rate: 54Mbps actual r is calculated with RBAR [3] used for BF Chair for Communication Networks, TU Chemnitz EUNICE 2011 Page 11
Routing Algorithm Frame Forwarding The path with minimum airtime metric is selected The transmission scheme is adapted with respect to the selected path Path Maintenance MUX Path Failure the node recognizing the failure generates PERR frames the node selects a path from the DPT and continues transmission with BF this avoids the loss of queued and on-air data frames the source node starts a new route discovery for OPT (O-PREQ) Directional Path Failure the node recognizing the failure generates PERR frames the source starts a new route discovery for OPT and DPT Chair for Communication Networks, TU Chemnitz EUNICE 2011 Page 12
Routing Algorithm - Overview Chair for Communication Networks, TU Chemnitz EUNICE 2011 Page 13
Routing Algorithm - Details Chair for Communication Networks, TU Chemnitz EUNICE 2011 Page 14
Routing Algorithm - Details Chair for Communication Networks, TU Chemnitz EUNICE 2011 Page 15
Simulation Results Scenario 1: 2 Node Topology S 50m D the source node S generates CBR traffic (5Mbps) the SNR of the link is decreased by 10dBm in the time interval 8-11s Results: Throughput O-MUX = HWMP using MUX Chair for Communication Networks, TU Chemnitz EUNICE 2011 Page 16
Simulation Results Scenario 1: 2 Node Topology Results: Delay Time Chair for Communication Networks, TU Chemnitz EUNICE 2011 Page 17
Simulation Results Scenario 2: Random Network Topology 30 nodes placed in a square field with variable edge length x 10 source-destination pairs, each source generates traffic at 100Kbps Results: Throughput Chair for Communication Networks, TU Chemnitz EUNICE 2011 Page 18
Conclusion Summary MHWMP improves the WMN throughput and shows a quite robust behavior against link outages caused by SNR degradations It exploits the advantages of smart antennas by adaptively using either spatial multiplexing or beamforming Future Work Investigate the performance of MHWMP in large scale wireless mesh network scenarios Extend MHWMP by a modified MAC layer to enable quality of service differentiation for traffic flows Chair for Communication Networks, TU Chemnitz EUNICE 2011 Page 19
References [1] Akyildiz, I., Wang, X.: A Survey on Wireless Mesh Networks. Communications Magazine, IEEE, vol. 43, pp.23 30. (2005) [2] Sundersasan, K.: Network Protocols for Ad-Hoc Networks with Smart Antennas. PhD Thesis, Georgia Institute of Technology, USA (2006) [3] Holland, G., Vaidya, N., and Bahl, P.: A Rate-Adaptive MAC Protocol for Multi-hop Wireless Networks. In: MobiCom 01, New York, NY, USA, pp. 236 251 (2001) [4] Alamouti, S. M.: A Simple Transmit Diversity Technique for Wireless Communications. IEEE J. Selecl. Areas Commun., 16(8), pp.1451 1458 (1998) [5] Wolniansky, P.W., Foschini, G.J., Golden, G.D., Valenzuela, R.A.: V-BLAST: An Architecture for Realizing Very High Data Rates over the Rich-Scattering Wireless Channel. In: ISSSE 98, pp. 295 300 (1998) [6] Takai, M., Martin, J., Bagrodia, R., Ren, A.: Directional Virtual Carrier Sensing for Directional Antennas in Mobile Adhoc Networks. In: 3rd ACM International Symposium on Mobile Adhoc Networking & Computing, MobiHoc 02, pp. 183 193, New York, NY, USA. (2002) Chair for Communication Networks, TU Chemnitz EUNICE 2011 Page 20
Thankyou Q? Chair for Communication Networks, TU Chemnitz EUNICE 2011 Page 21