Analyzing Multi-hop MIMO Network Throughput: Spatial Reuse, Multiplexing, and Interference, Slides of Computer Networks

Download Analyzing Multi-hop MIMO Network Throughput: Spatial Reuse, Multiplexing, and Interference and more Slides Computer Networks in PDF only on Docsity! Characterization and analysis of multi-hop wireless mimo network throughput 1 Docsity.com Overview • Introduction • MIMO Basics • Problem Statement – Protocol descriptions • Packet-level Constraints • Test Setup • Results • Conclusions 2 Docsity.com MIMO Basics • Transmitter and receiver both have multiple physical antennas 5 Docsity.com MIMO Basics • Transmitter uses a weighting vector u = [u1, u2] while receiver uses a weighting vector v = [v1, v2] • Channel coefficient matrix H (??) • s(t) = transmitting signal, r(t) = received signal • r(t) = (uHv) * s(t) • We can find values for u and v such that uHv at target = 1 and at other nodes = 0 6 Docsity.com MIMO Basics ¢ Multiple-stream signals 8 ,(t) 7 MIMO Basics: Spatial Reuse • Node 4 can receive transmission from 3 with spatial reuse • Find v = [v1, v2] such that (u2H2,4)v = 1 and (u 1 H1,4)v = 0 10 Docsity.com MIMO Basics: Multiplexing • Node 4 could instead use its antennas to receive two streams from node 2 • Constraints outlined before: use two weight vectors such that – r(t) 1 gets s(t) 1 at full strength – r(t) 2 gets s(t) 1 at zero strength – r(t) 1 gets s(t) 2 at zero strength – r(t) 2 gets s(t) 2 at full strength 11 Docsity.com Interference Avoidance Models • Ways to ensure that the constraints outlined so far can be followed • Non-cooperative Interference Avoidance (NiM) – Transmitters find weight vectors to null their signal at all receivers before transmitting – Receivers find weight vectors to null their signal from all nearby transmitters before receiving • Cooperative Interference Avoidance (CiM) – Either the transmitter OR the receiver ensures that no interference takes place (solve system for u OR v) 12 Docsity.com Characterization and analysis of multi-hop wireless mimo network throughput 1 Docsity.com Overview • Introduction • MIMO Basics • Problem Statement – Protocol descriptions • Packet-level Constraints • Test Setup • Results • Conclusions 2 Docsity.com MIMO Basics • Transmitter and receiver both have multiple physical antennas 5 Docsity.com MIMO Basics • Transmitter uses a weighting vector u = [u1, u2] while receiver uses a weighting vector v = [v1, v2] • Channel coefficient matrix H (??) • s(t) = transmitting signal, r(t) = received signal • r(t) = (uHv) * s(t) • We can find values for u and v such that uHv at target = 1 and at other nodes = 0 6 Docsity.com MIMO Basics ¢ Multiple-stream signals 8 ,(t) 7 MIMO Basics: Spatial Reuse • Node 4 can receive transmission from 3 with spatial reuse • Find v = [v1, v2] such that (u2H2,4)v = 1 and (u 1 H1,4)v = 0 10 Docsity.com MIMO Basics: Multiplexing • Node 4 could instead use its antennas to receive two streams from node 2 • Constraints outlined before: use two weight vectors such that – r(t) 1 gets s(t) 1 at full strength – r(t) 2 gets s(t) 1 at zero strength – r(t) 1 gets s(t) 2 at zero strength – r(t) 2 gets s(t) 2 at full strength 11 Docsity.com Interference Avoidance Models • Ways to ensure that the constraints outlined so far can be followed • Non-cooperative Interference Avoidance (NiM) – Transmitters find weight vectors to null their signal at all receivers before transmitting – Receivers find weight vectors to null their signal from all nearby transmitters before receiving • Cooperative Interference Avoidance (CiM) – Either the transmitter OR the receiver ensures that no interference takes place (solve system for u OR v) 12 Docsity.com Problem Statement • Network Model • L = set of all node pairs (m,n) such that m can transmit to n (individual links referred to as i) • Lm+ : set of all links whose transmitter is m • Lm- : set of all links whose receiver is m • Lm : Lm+ U Lm- 15 Docsity.com Problem Statement • C = set of all link pairs (i,j) such that a transmission on i will interfere with a transmission on j • Ci+ : set of all links whose receivers interfere with i’s transmission • Ci- : set of all links whose transmitters interfere with i’s reception 16 Docsity.com Protocols • Spatial Reuse Only MIMO Protocol (SRP) – All of a node’s degrees of freedom are dedicated to preventing interference and increasing spatial reuse • Spatial Multiplexing Only MIMO Protocol (SMP) – All of a node’s degrees of freedom are dedicated to transmitting and receiving additional simultaneous streams 17 Docsity.com Constraints: SRP • Only one link per node may be active at a time • Under NiM, every sender or receiver must ensure it has enough degrees of freedom • When attempting to transmit, the above simplifies to Σ yj + 1 ≤ β (or α) 20 Docsity.com Constraints: SRP • Under CiM, only one of the transmitter or receiver needs to null the signal • Constraint defined by sum of nulled signals rather than individual degrees of freedom 21 Docsity.com Constraints: SMP • zi : number of active streams over link i • Still only one active link at a time • No spatial reuse, so only one active link for every contending pair 22 Docsity.com LP Relaxation • Relax constraints from instantaneous to average • yi = • Same for λ, μ, z, θ, υ. • Relaxed constraints same format as instantaneous, but use the average over a time slot set S 25 Docsity.com Test Setup • Network Parameters – Link capacity set to 1 unit/second – Degrees of freedom strictly equal to antennas – 100mx100m space with random distribution – Q total source-destination pairs (active flows) • Transmission Range – Controls degree of nodes and interference • Node Density – Increases node degree, but not interference • Hop Length I h f i f 26 Docsity.com Results: SRP • Asymptotic bound – Once all medium contention has been resolved, no further improvement can be made • High transmission ranges suffer too much interference at low antenna numbers, but when more antennas are added, the benefits of increased node degree help it perform 27 Docsity.com Results: SMP • Transmission Range – Increasing interference hurts throughput • Node Density – Not explained – slight interference issues? • Hop Length – Worse (increased contention/interference) 30 Docsity.com Results: SRMP • No asymptotic bound – SRMP resolves medium conflicts, but because extra degrees of freedom can be used for multiplexing, additional antennas continue to increase throughput 31 Docsity.com Results: SRMP • Transmission Range – Unique maximums for each antenna setup • Node Density – Increasing node degree increases throughput • Hop Length – Worse (increased contention/interference) 32 Docsity.com Results: NiM v. CiM • CiM always outperforms – Additional degrees of freedom available for more reuse or multiplexing 35 Docsity.com Conclusions • LP problem of optimal throughput over a MIMO network solved under various configurations • MIMO protocols and interference models can be used by network designers 36 Docsity.com Conclusions • A lot of future research to be done • Many actual implementation decisions and problems are left to future work or not even mentioned – Discovery of u, v, H – Heterogeneous networks – Cooperation mechanism for CiM 37 Docsity.com