SLIDE 1
Effective Capacity Through Physical and Data-Link Layers Sami Akin - - PowerPoint PPT Presentation
Effective Capacity Through Physical and Data-Link Layers Sami Akin - - PowerPoint PPT Presentation
Effective Capacity Through Physical and Data-Link Layers Sami Akin Institute of Communications Technology Leibniz Universitt Hannover 1 / 14 Outline Background and Motivation Effective Capacity Cognitive Radio Concerns and Analyses
SLIDE 2
SLIDE 3
Background and Motivation
In Wireless communications,
◮ The principal design problem in the past:
◮ Time-varying and frequency-selective propagation path ◮ Gaussian noise
◮ Contemporary problems:
◮ Spectrum scarcity ◮ Interference ◮ Power consumption
◮ Change in the objective:
◮ Capacity maximization limited by fading and noise (past) ◮ Network capability, e.g., delay and backlog (recent)
3 / 14
SLIDE 4
Cross-Layer Concerns
Figure: Conventional OSI Figure: Prospective OSI 4 / 14
SLIDE 5
Effective Capacity
Figure: A system with a known stochastic service process s(t)
◮ For a stable system, a(t) =? ◮ Effective Capacity
◮ Dual of Effective Bandwidth ◮ Maximum constant arrival rate a stochastic service process
can sustain under certain QoS constraints specified by θ
EC(θ) = − lim
t→∞
1 tθ loge E
- e−θ t
τ=1 s(τ)
5 / 14
SLIDE 6
What to infer from θ?
Figure: Queue in steady-state
θ = − lim
q→∞
log Pr{Q > q} q
◮ For large q: Pr{Q > q} ≈ e−θq ◮ Larger θ → stricter constraints on buffer ◮ Smaller θ → looser constraints on buffer ◮ Properties of Effective Capacity:
◮ limθ→0 EC(θ) =
⇒ average service rate
◮ limθ→∞ EC(θ) =
⇒ minimum service rate 6 / 14
SLIDE 7
Cognitive Radio
A communications model allowing unlicensed (secondary) users to operate in the spectrum with the presence of licensed (primary) users
◮ Access strategies:
◮ Interweave (Channel sensing required) ◮ Underlay (Interference power limitation) ◮ Overlay (Cooperation with licensed users required)
◮ A hybrid strategy of Interweave and Underlay:
First: Channel sensing Second: Power level adjustment Third: Transmission
7 / 14
SLIDE 8
Cognitive Radio Framework
Regarding channel sensing decision and its correctness:
- 1. Channel is busy, and detected as busy (correct sensing)
- 2. Channel is busy, and detected as idle (miss-detection)
- 3. Channel is idle, and detected as busy (false alarm)
- 4. Channel is idle, and detected as idle (correct sensing)
8 / 14
SLIDE 9
Physical Layer Transmission Framework
9 / 14
SLIDE 10
Constraints on Cognitive Radios
◮ Channel sensing with errors: False alarms and
miss-detections
◮ Strictly limited transmission power levels ◮ Transmission rates that depend on channel sensing results ◮ Increased number of transmission outages ◮ Decreased data transmission rates
10 / 14
SLIDE 11
Throughput Analysis
◮ Case I: Channel fading is known at the receiver
⋆ Data is forwarded at constant rates depending on channel sensing results ⋆ Transmission outages occur when the rates are greater than the instantaneous channel capacity ⋆ Objective: Rates that maximize the effective capacity under given constraints
◮ Case II: Channel fading is known at both the transmitter
and the receiver
⋆ Data is forwarded at rates equal to the capacity regarding channel sensing results ⋆ Transmission outages due to miss-detections ⋆ Objective: Effective capacity performance with channel sensing errors under different channel conditions 11 / 14
SLIDE 12
Design Concerns
◮ Transmission outages
⋆ When to (not to) take the risk of transmission outages
◮ Channel sensing
⋆ The interplay between the sensing quality and its duration
◮ Input distribution
⋆ Performance investigation with arbitrary input distributions rather than Gaussian distributed signals
◮ Channel estimation
⋆ Imperfect channel estimation results ⋆ Uni-directional effect of channel sensing on channel estimation performance
◮ Channel encoding/decoding
◮ Encoding and decoding performance with different
techniques 12 / 14
SLIDE 13
Asymptotic Analysis
Rather than the effective capacity expression, we can utilize certain approximations:
◮ Low signal-to-noise ratio (SNR) regime
⋆ Approximation of the effective capacity when SNR goes to zero ⋆ Good for energy-per-bit investigations
◮ High SNR regime
⋆ Approximation of the effective capacity when SNR goes to infinity ⋆ Good for the analysis of the effects of transmission outages 13 / 14
SLIDE 14