Sen Probe Path Capacity Estimation in Wireless Sensor

Sen. Probe: Path Capacity Estimation in Wireless Sensor Networks Tony Sun, Ling-Jyh Chen, Guang Yang M. Y. Sanadidi, Mario Gerla Sen. Metrics 05

Outline n n Motivation Background q q n n n Design Consideration What do we want to measure? Effective Capacity Previous Method Proposed Approach: Sen. Probe Simulation Results Conclusion Sen. Metrics 05 2

Motivation n Mobile computing platforms may interact with ambient sensor environment establishing “Opportunistic wireless networks” n Evaluation and measurement of wireless path capacity in sensor network is of realistic interest q n (i. e. Capacity planning, protocol design, performance analysis, system deployment, assess applicability of deployment) Need a tool that can monitor and measure opportunistic path capacity in wireless sensor networks Sen. Metrics 05 3

Outline n n Motivation Background q q n n n Design Consideration What does Sen. Probe actually measure? Effective Capacity Previous Method Proposed Approach: Sen. Probe Simulation Results Conclusion Sen. Metrics 05 4

Design Consideration n CSMA-CA and variant schemes still popularly used in sensor network for its simplicity q n (IEEE 802. 15. 4 beaconless mode, Berkeley motes, etc) Basic CSMA-CA doesn’t incorporate RTS/CTS mechanisms q q q Send packet when an idle channel is detected Smaller packet overhead if idle channel can be detected quickly Suffers from hidden terminal problem Sen. Metrics 05 5

What do we want to measure? (1) n The effective end-to-end rate is defined as the maximum achievable data rate in the absence of any cross traffic connection. n It is smaller than the raw data rate at the physical layer due to q q Packet Overhead Interference between multiple packets in the pipeline Sen. Metrics 05 7

What do we want to measure? (2) n In fact, path capacity in wireless net also varies with: q q q q MAC protocol and link scheduling Link interference S/N ratio; Tx power Encoding/modulation scheme Number of antennas (eg MIMO) Antenna directionality etc Sen. Metrics 05 8

Neighborhood Example n n If Dr=Di=250 m , nodes {3, 4, 5} are within the same n-hood, C’=C/3 If Dr=250 m, Di=500 m, nodes {2, 3, 4, 5, 6} are in n-hood, C’=C/4 Dr= effective receive range from node 4 (solid-line circle) Di = interference range caused by node 4 (dotted-line circle) Distance between nodes: 200 m Sen. Metrics 05 9

Effective Capacity of CSMA-CA Enabled Wireless Channel n The effective capacity of a one-hop link can be calculated as n For the CSMA environment in our study (if ACKs are used) Sen. Metrics 05 10

Previous Work (Morris et al) n n Dr=250 m, Di=500 m Use UDP flows to probe the maximum achievable throughput (brute force method) Sen. Metrics 05 11

Outline n n Motivation Background q q n n n Design Consideration What does Sen. Probe actually measure? Effective Capacity Previous Method Proposed Approach: Sen. Probe Simulation Results Conclusion Sen. Metrics 05 12

Cap. Probe Concept n Key insight: a packet pair that gets through with zero queueing delay yields the exact estimate Capacity Sen. Metrics 05 13

Issues: Compression and Expansion • Queueing delay on the first packet => compression • Queueing delay on the second packet => expansion Sen. Metrics 05 14

Sen. Probe n Path capacity estimation tool specially designed for the multi-hop CSMA based wireless networks. q q One-way estimation technique, based on Cap. Probe concepts Aimed to simplify the path capacity estimation process n A back-to-back packet train technique designed to overcome the hidden terminal effects in CSMA environment n Sen. Probe measures end-to-end effective capacity in wireless ad hoc networks. n Sen. Probe is simple, fast and less intrusive to comparative techniques. Sen. Metrics 05 15

Sen. Probe Algorithm(1) n Instead of using back-to-back packet pairs, Sen. Probe relies on back-to-back packet train to overcome the effect of hidden terminal in CSMA-CA n The length of this back-to-back packet train depends on the interference range and the transmission range of the specific radio technology under question Sen. Metrics 05 16

Sen. Probe Algorithm(2) n The receiver measures the OWD of every packet in kth packet train received as the difference between time received and time sent n the minimum OWDSUM is kept for the kth packet train. The “good” dispersion sample r (i. e. samples encountering no cross traffic) is the sample with the minimum OWD sum n Dispersion of the good sample calculated, and used to estimation capacity Sen. Metrics 05 17

Sen. Probe-Visualization 1) 2) 3) 4) Sen. Metrics 05 18

Outline n n Motivation Background q q n n n Design Consideration What does Sen. Probe actually measure? Effective Capacity Previous Method Proposed Approach: Sen. Probe Simulation Results Conclusion Sen. Metrics 05 19

Simulation Results (1) Path Capacity measured via FTP connection and Packet-Pair technique (one way Cap. Probe) Sen. Metrics 05 20

Simulation Results (2) Path capacity of adhoc multi-hop forwarding chain in CSMA-CA wireless environment Sen. Metrics 05 21

Simulation Results (3) End-to-end capacity estimation of multi-hop connections within the same collision domain Sen. Metrics 05 22

Simulation Results (4) Capacity estimates along a multi-hop forwarding chain for CSMA-CA with ACK enabled wireless sensor network Sen. Metrics 05 23

Conclusion n Sen. Probe uses back-to-back packet trains, and relies on packet dispersion between the packet trains to measure the path capacities in a one-way fashion. n Sen. Probe estimates e 2 e path capacity in CSMA enabled wireless sensor networks. n Sen. Probe is a simple and non-intrusive technique that can accurately reflects the effective path capacity Sen. Metrics 05 25

Thanks! Sen. Metrics 05 26