“WOMEN” Project Meeting Rome, Jan. 19th, UNITN Research Unit Status of the Work – First Year Fabrizio Granelli DIT – University of Trento

“WOMEN” Project Meeting Rome, Jan. 19th, UNITN role in the project Goal: “Performance Enhancement of Wireless Mesh Networks and Testbed Implementation” WP1 WP3  T3.4 Stream Control WP4  T4.5 Network Capacity Analysis WP5 (Resp.)  T5.6 (Testbed)(Resp.)  T5.7 (Trials)(Resp.) Ongoing activities

“WOMEN” Project Meeting Rome, Jan. 19th, To define and validate an analytical model for connectivity and capacity of Wireless Mesh Networks, given design parameters n, r, l. Traffic from/to mesh base station ≠ ad-hoc scenario Network Capacity Analysis: goal E. Miorando, F. Granelli, “On Connectivity and Capacity of Wireless Mesh Networks”, ICC 2007, accepted

“WOMEN” Project Meeting Rome, Jan. 19th, Problem Statement n→ # nodes (random distribution) S→ squared area of dimension l x l m → # access points r → comm range n n → # neighbor nodes n f → # forwarding nodes

“WOMEN” Project Meeting Rome, Jan. 19th, Problem Statement Single-cell analysis → squared area → m = 1 → interfering nodes Traffic to/from BS Hyp.: → IEEE CSMA/CA – RTS/CTS * → MFR routing → r = 250 [m] * → “Performance Analysis of the IEEE Distributed Coordination Function”, by G. Bianchi (JSAC, March 2000)

“WOMEN” Project Meeting Rome, Jan. 19th, Connectivity analysis No. of forwarding nodes * Poisson distribution: λ = n f * → “Optimal Transmission Ranges for Randomly Distributed Packet Radio Terminals”, by Takagi e Kleinrock (March 1984)

“WOMEN” Project Meeting Rome, Jan. 19th, Connectivity analysis

“WOMEN” Project Meeting Rome, Jan. 19th, Bottleneck * Bottleneck collision domain * Capacity analysis GGGGGGGG G2G3G4G5G6G7G8G * → “The Nominal Capacity of Wireless Mesh Networks”, by Jun e Sichitiu (Oct. 2003)

“WOMEN” Project Meeting Rome, Jan. 19th, Capacity Analysis n i →connected nodes at i-hops distance from the BS

“WOMEN” Project Meeting Rome, Jan. 19th, Validation of the Approach NS-2 simulator (v. 2.29) → C++ e OTcl No WMN support → set tcp [new Agent/TCP] $tcp set class_ 2 set sink [new Agent/TCPSink] $ns_ attach-agent $node_(11) $tcp $ns_ attach-agent $node_(19) $sink $ns_ connect $tcp $sink set ftp [new Application/FTP] $ftp attach-agent $tcp $ns_ at "$ftp start"

“WOMEN” Project Meeting Rome, Jan. 19th, Experimental Results: p conn Averaged on 50 runs variable n l = 2500 r = 250 Max error: 2,25 % Avg. stdev: 8,15 %

“WOMEN” Project Meeting Rome, Jan. 19th, Experimental Results: Capacity Averaged on 50 runs variable n l = 2500 r = 250 Avg. error: 17,53 % Avg. stdev: 22,85 % [kbit/sec]

“WOMEN” Project Meeting Rome, Jan. 19th, Averaged on 20 runs variable n l = 2500 r = 250 Avg. error: 6,37 % Avg. stdev: 17,55 % Averaged on 20 runs variable n l = 2500 r = 250 Avg. error: 15,51 % Avg. stdev: 22,15 % Experimental Results: Capacity [kbit/sec]

“WOMEN” Project Meeting Rome, Jan. 19th, Conclusions A first model for connectivity and capacity of WMNs was developed and validated Good accuracy Points to consider:  Inter-cell interference  Heterogeneous WMN (e.g. WiMAX+WiFi)  Node movement (statistical model)

“WOMEN” Project Meeting Rome, Jan. 19th, UNITN / CREATE-NET testbed

“WOMEN” Project Meeting Rome, Jan. 19th, Testbed: where are we? Started from Roofnet, making it device- agnostic:  Monitor mode  ad hoc mode  Link quality evaluation using ETX Roofnet on laptops works!  Roofnet patch available on WOMEN website

“WOMEN” Project Meeting Rome, Jan. 19th, Testbed: where are we? Software porting to openwrt successful Automatic Gateway Discovery working Meraki (Roofnet spin-off) openwrt platform freely available  Click in userspace running on 200MHz CPUs  working in b mode

“WOMEN” Project Meeting Rome, Jan. 19th, 2007 Click modules:  Read RSSI, noise, bitrate, etc. from incoming packet  Translate ingoing IEEE packet into a Ethernet frame  Set rate and transmission power of the outgoing packet Preamble management in Click

“WOMEN” Project Meeting Rome, Jan. 19th, 2007  Change from monitor mode to ad-hoc mode (supported by any IEEE NIC)  Introduction of an element for static setting of preamble parameters  link-quality evaluation using ETX flags = 0; power = 60; rssi = 8; rate = 2; retries = 0; “BlankDecap” module settings Roofnet Modifications

“WOMEN” Project Meeting Rome, Jan. 19th, 2007 SCHEDULER prototypes First prototype - COMPLETED:  Priority scheduling based (WRR)  Works for TCP/UDP (2 hops test) Second prototype – WORK IN PROGRESS:  TRADE-OFF UTILIZATION/FAIRNESS  Priority scheduling based (WRR)  RED queuing policies (avoid drop-tail) ?  Fairness idea: use deficit queues (one for each service class) which provides FQ fairness  Channel awareness ? Channel measurements -> long implementation proper temporal scale to avoid fading and multipath phenomena

“WOMEN” Project Meeting Rome, Jan. 19th, 2007 Measurement platform (completed) Set of bash scripts for:  Automatic ssh key distribution among nodes  Tcp and udp bi-directional distributed measurements (iperf) Throughput Jitter delay  Results collection and post-processing  Can launch arbitrary unix commands on remote nodes

“WOMEN” Project Meeting Rome, Jan. 19th, Dissemination activities - Papers E. Miorando, and F. Granelli "On Connectivity and Capacity of Wireless Mesh Networks,“ IEEE International Conference on Communications (ICC’07), Glasgow, Scotland, June 2007 (accepted). D. Kliazovich, F. Granelli and D. Miorandi “Logarithmic Window Increase for TCP Westwood+ Improvement in High Speed, Long Distance Networks,” Computer Networks (submitted).

“WOMEN” Project Meeting Rome, Jan. 19th, UNITN Research Unit Future Activities Fabrizio Granelli DIT – University of Trento

“WOMEN” Project Meeting Rome, Jan. 19th, Activities in the 2nd Year Finalization of Network Capacity Analysis Testbed Design:  + WiMAX-like long distance link  Frequency Testbed Implementation:  Second prototype design and programming  Testbed deployment  Tests on testbed  Results collection and analysis