Receiver-Initiated Channel Hopping (RICH) Makis Tzamaloukas Computer and Communications Research Group (CCRG) Computer Engineering Department Jack Baskin School of Engineering University of California Santa Cruz, CA 95064
February 9th, UCSC A. E. Tzamaloukas2 Presentation Outline n Introduction n Physical Layer n Motivation n Polling Issues n RICH Ô RICH-SP Ô RICH-DP n Throughput Analysis n Delay Analysis n Simulations n Conclusions
February 9th, UCSC A. E. Tzamaloukas3 Physical Layer - Unlicensed RF n FCC regulations require the use of frequency hopping (FH) or direct sequence (DS) spread spectrum modulation to operate in an ISM band n Multiple users can share the available bandwidth at the same time at a minimal increase of complexity and cost
February 9th, UCSC A. E. Tzamaloukas4 Physical Layer - FHSS n Frequency Hopping Spread Spectrum (FHSS) In the USA the: 915 MHz band has 52 FH channels 2.4 GHz band has 79 FH channels 5.8 GHz band has 125 FH channels On the right, two pairs of nodes exchange DATA packets by following a unique hopping pattern
February 9th, UCSC A. E. Tzamaloukas5 Physical Layer - FHSS n Hopping sequence: The pattern with which nodes use the channels n Advantages: robustness against multi-path propagation, minimize hidden node terminal problems, increased security, not prone to fading, capable to capture a packet even when multiple packets overlap n Most commercially available ISM radios are FH
February 9th, UCSC A. E. Tzamaloukas6 Motivation n The receiver of a data packet is the point of interest n Recast the collision avoidance dialogues so that the receiver, sender or both can have control of the dialogue n Provide correct floor acquisition without carrier sensing and code assignment n Be applicable to multi-channel frequency-hopping or direct- sequence spread-spectrum radios
February 9th, UCSC A. E. Tzamaloukas7 Polling Issues n When to poll: whether or not the polling rate is independent of the data rate at polling nodes Ô independent polling Ô data-driven polling n To whom: whether the poll is sent to a particular neighbor or to all neighbors; for dense networks a schedule may have to be provided to the poll recipients n How: whether the polling packet asks for permission to transmit as well
February 9th, UCSC A. E. Tzamaloukas8 RICH Characteristics n Dwell time should be long enough to transmit a pair of MAC addresses, a CRC and framing bits n Use synchronous frequency hopping to ensure that all radios hop to different frequency hops at the same time n Nodes do not need carrier sensing or code assignment n Commercially available radios can be used
February 9th, UCSC A. E. Tzamaloukas9 RICH-SP h1 h2 h3 h4 t1t3t2t4t5t6 hop time All the nodes follow a common channel-hopping sequence. If a node receives an RTR then it sends its data to the polling node over the same channel hop; all the other nodes hop to the next channel hop. RTR DATARTRsilenceDATACTSRTR backoff RTR
February 9th, UCSC A. E. Tzamaloukas10 RICH-DP n The key difference from RICH-SP is that now an RTR is an invitation to receive and transmit; therefore, two data packets can be exchanged in the same busy period h1 h2 h3 h4 t1t3t2t4t5t6 hop time t6 RTRsilence DATACTSRTR backoff RTR DATA
February 9th, UCSC A. E. Tzamaloukas11 Throughput Analysis Model n ad-hoc network of N nodes n multiple channels, error-free n the size of an RTR and CTS is less than one slot; the size for a data packet is derived from a geometric pdf n the turn-around time is considered to be part of the duration of control and data packet n a polled node receiving an RTR always has a data packet to send n the probability that the packet is addressed to the polling node is 1/N n Analysis is based on a model first introduced by Sousa and Silvester [Trans. On Communications - March 1988]
February 9th, UCSC A. E. Tzamaloukas MACA-CT --- RICH-SP Fixed packet length Throughput analysis results n Throughput vs. probability of transmission. Results for RICH-SP are compared against MACA-CT [Joa-Ng and Lu - INFOCOM 1999] --- RICH-SP --- MACA-CT Fixed number of nodes
February 9th, UCSC A. E. Tzamaloukas13 Throughput analysis results n Throughput vs. probability of transmission. The packet length is fixed equal to 10 hops and the number of nodes in the network is a parameter. Results for RICH-DP are compared against RICH-SP.
February 9th, UCSC A. E. Tzamaloukas14 Delay analysis results Normalized delayActual delay
February 9th, UCSC A. E. Tzamaloukas15 Network Topologies Base N1 N2 B1 B2 N1 (a) (b) (c) Base
February 9th, UCSC A. E. Tzamaloukas16 Simulated Radio Model n Radio features – 2.4GHz FHSS, no capture, no power control – 80 channels, 1Mbps each – 120us dwell time – half-duplex operation – RICH MAC protocol – omni-directional antenna
February 9th, UCSC A. E. Tzamaloukas17 Simulation Results
February 9th, UCSC A. E. Tzamaloukas18 Simulation Results aggregate data rate < available bandwidth
February 9th, UCSC A. E. Tzamaloukas19 Simulation Results aggregate data rate > available bandwidth
February 9th, UCSC A. E. Tzamaloukas20 Simulation Results Comparison
February 9th, UCSC A. E. Tzamaloukas21 Conclusions n By reversing the collision avoidance handshake we improved the performance of MAC protocols for ad-hoc networks n RICH protocols achieve correct floor acquisition without carrier sensing or code assignment n RICH outperforms any other multi-channel collision avoidance MAC protocol to date in terms of throughput and delay n Extensive simulations verified our analytical results