User Cooperation for Improving Spatial Spectral Utilization in Multhop Wireless Networks March 1, 2006 Chansu Yu 3/1/2006 NEONet 2006
Overview Wireless multihop networks What’s wrong? Solutions for broadcast Solutions for packet relaying Solutions for spatial reuse Conclusions & future work 3/1/2006 NEONet 2006
Multihop Ad Hoc Networks Introduction Multihop Ad Hoc Networks 3/1/2006 NEONet 2006
HQ Mobile nodes, Wireless links Infrastructure-less Multi-hop routing Minimal administration HQ 3/1/2006 NEONet 2006
Introduction New serious applications Reconsiderations Wireless sensor networks Wireless mesh networks Reconsiderations 802.11 PHY/MAC Ad hoc routing algorithms Akyildiz, I.F., Wang, X. and Wang, W., “Wireless Mesh Networks: A Survey,” Computer Networks Journal (Elsevier), March 2005. 3/1/2006 NEONet 2006
What’s Wrong? 802.11 PHY/MAC is optimized for single-hop communication “Relay” not considered the most frequent operation in multihop scenarios CS, RTS/CTS and NAV is to provide an exclusive access to the medium because every communication involves the AP Spatial reusability is important Broadcasting is meant to be one-hop Broadcast storm problem in multihop scenarios Shortest-path routing algorithms are not practical Expected transmission time (ETT), … 3/1/2006 NEONet 2006
Solutions for Broadcast Gossip (INFOCOM ‘02) Manycast (MobiHoc ’03) Epidemic routing (Computer, May ‘04) RandomCast (ICDCS ’05) Unconditional broadcast S D A B C RandomCast Who to relay? 3/1/2006 NEONet 2006
Solutions for Packet Relaying Packet salvaging at network layer “Packet salvaging” in DSR “Local repair” in AODV Packet salvaging at MAC layer Extremely Opportunistic Routing (HotNet-II’03) Implicit Geographic Forwarding (TR, UVA, ‘03) Geographic Random Forwarding (TMC, April’03) Multiple Access with Salvation Army (MobiHoc’05) 3/1/2006 NEONet 2006
MASA Algorithm No collision Collision at receiver s Salvation army i j (1) DATA (1) DATA (1) DATA (3) SDATA (2) ACK (2) SACK (4) ACK Salvation army 3/1/2006 NEONet 2006
MASA: Selection of Relay Node Based on “relaying backoff” A candidate gives up its relaying activity when the packet is relayed by another candidate Otherwise, it becomes the relay node and sends SACK Sender (i) Receiver (j) Relay (s) DATA D ACKTimeout DIFS SIFS D ACK tr. time ACK (not delivered) t SACK Relaying backoff ( t S ) Relaying interval (T SI ) Who to relay? 3/1/2006 NEONet 2006
Simulation Environment System parameters Simulation time: 900s 100 nodes in 3001500 m2 Mobility parameters Random waypoint mobility model Maximum speed: 5m/s Simulator ns2 2.27 with modification for cumulative interference and signal capture Traffic parameters 40 CBR connections Packet rate: 3 packets/s Packet size: 256 bytes 40 — 100 TCP connections Other parameters AODV routing Two-ray ground propagation model 250m TR range; 350m and 550m CS range for MASA and DCF, respectively 3/1/2006 NEONet 2006
Simulation with A Single Interferer Simulation scenario Simulation parameters Simulation time: 180s Transmission range: 250m Propagation model: two-ray ground propagation channel Packets: 512 bytes CBR or TCP packets Data rate: 2Mbps (a) Direct scenario Major interest Interfering communication (b) Relaying scenario Relay node 3/1/2006 NEONet 2006
Simulation Results with CBR Traffic (a) Direct scenario (b) Relaying scenario Instantaneous throughput (kbps) Instantaneous throughput ( kbps) Packet relaying is helpful to alleviate the unfairness problem and to improve the performance. Unfairness for i-j communication Fair and Higher aggregate throughput 3/1/2006 NEONet 2006
Simulation with Multiple Interferers DCF2 DCF with two-way handshake (without RTS/CTS) DCF4 DCF with four-way handshake (with RTS/CTS) MASA With two-way handshake (without RTS/CTS) 3/1/2006 NEONet 2006
Solutions for Spatial Reuse Proactively look for an opportunity to transmit my packet when I’m not relaying Let us consider the radio propagation model 3/1/2006 NEONet 2006
Radio Propagation Model Two-ray ground propagation model Reception Model CPj CSj i j TRj (z0: capture ratio) i j 3/1/2006 NEONet 2006
Radio Propagation Model Two-ray ground propagation model Reception Model CPj CSj i j TRj (z0: capture ratio) i j Nodes in this area do not cause collisions 3/1/2006 NEONet 2006
DCF of IEEE 802.11 Carrier sense mechanism Hidden terminal problem Vulnerable Space (VS) Nodes within VS can not sense ongoing communication but can cause collision to receiver. Exposed terminal problem Wasted Space (WS) Nodes within WS can sense ongoing communication but will not cause collision to receiver. 3/1/2006 NEONet 2006
DCF of IEEE 802.11 Carrier sense mechanism Hidden terminal problem Vulnerable Space (VS) Nodes within VS can not sense ongoing communication but can cause collision to receiver. Exposed terminal problem Wasted Space (WS) Nodes within WS can sense ongoing communication but will not cause collision to receiver. 3/1/2006 NEONet 2006
An Example of VS and WS 915 MHz WaveLAN radio hardware Transmission range 250m Carrier sense range 550m Communication distance 200m Capture zone 356m (when z0=10dB) VS ≈ WS = large 3/1/2006 NEONet 2006
Solutions for Spatial Reuse Carrier sense mechanism in this case not only reduces interference but also solves the hidden terminal problem (VS becomes very small). But it introduces another serious problem, the exposed terminal problem (large WS), which greatly reduces the spatial reusability. CS, RTS/CTS, EIFS in 802.11 fail to efficiently reuse the spatial spectral resource. 3/1/2006 NEONet 2006
Solutions for Spatial Reuse Adaptive Selection of Carrier Sense Thresholds (TR, UIUC, ‘04): product of transmit power and CS threshold should be kept to be a fixed constant Adaptive Physical Carrier Sensing (ICC’04) Optimal Physical Carrier Sensing (INFOCOM ’05) Physical Carrier Sensing and Spatial Reuse (INFOCOM ‘06) Aggressive Virtual Carrier Sensing (Globecom’03) Enhanced Carrier Sensing (IFIP Networking, ’04): Adaptive EIFS scheme Is it safe to transmit if not relaying? 3/1/2006 NEONet 2006
Future Work Who to Relay & When to Tx? R. Ramanathan, “A Radically New Architecture for Next Generation Mobile Ad Hoc Networks,” ACM Mobicom, 2005. 3/1/2006 NEONet 2006
Another Dimension: Cooperative Communication ? Destination When a packet is not received at the Destination (or received with error), a relay node that successfully overheard the packet may relay this to the destination. Source Relay 3/1/2006 NEONet 2006 Klaus Fosmark, University of Texas at Dallas, 3/11/05
A. Nosratinia, T. Hunter, A. Hedayat, Cooperative communication in wireless networks, IEEE Comm., 2004 3/1/2006 NEONet 2006
Approaches Move “Routing” and “Forwarding” functions to the physical layer Extract certain information (destination, signal strength,..) from the front of the packet Define new PHY format Neighborhood information is critical 3/1/2006 NEONet 2006
THANK YOU ! 3/1/2006 NEONet 2006