Outage Probability & Diversity Order Analysis

• Outage probability quantifies the likelihood that a wireless channel’s capacity falls below a target rate, while diversity order defines the high-SNR decay rate of this probability.
• The topic covers diverse mechanisms such as MRC, block-fading, cooperative relaying, and HARQ, demonstrating their impact on achieving performance gains.
• Insights from the analysis guide system design trade-offs including precoding, rate selection, and resource allocation to optimize reliability in interference-limited networks.

Outage probability is a central metric in the analysis of wireless and cooperative communication systems, quantifying the probability that a system’s instantaneous channel capacity falls below a target rate, given random fading or interference. Diversity order is a fundamental performance parameter expressing the asymptotic rate at which the outage probability decays as the SNR (or a relevant network parameter) increases. Together, outage probability and diversity order form the theoretical cornerstone for the robust design and comparison of transmission schemes in fading, interference-limited, and cooperative networks.

1. Formal Definitions and General Principles

The outage probability, $P_{\mathrm{out}}$, for a target rate $R$ and SNR $\rho$ is defined as

$$P_{\mathrm{out}}(\rho,R) = \mathbb{P}\{ I(\text{channel};\rho) < R\}$$

where $I(\cdot)$ is the instantaneous mutual information, and the probability is taken over the channel, fading, and possibly interference realizations (Jabi et al., 2011, 0710.1595, Choi, 2020).

Diversity order $d$ characterizes the slope of the outage probability in the high-SNR regime: $$d = -\lim_{\rho\to\infty} \frac{\log P_{\mathrm{out}}(\rho)}{\log \rho}$$ A diversity order $d$ implies that $P_{\mathrm{out}}(\rho)\sim C\rho^{-d}$ for large $\rho$, where $R$0 encodes coding gain, channel statistics, and protocol specifics (Choi, 2020, Jabi et al., 2011, Zhong et al., 2012).

2. Diversity Mechanisms and Prototypical Outage Expressions

Diversity can be achieved temporally, spatially, in frequency, or through ARQ/HARQ retransmission or cooperative means. The prototypical example is $R$1-branch Maximal Ratio Combining (MRC) in Rayleigh channels: the post-combining SNR is a sum of independent exponentials, whose CDF gives outage

$R$2

yielding $R$3 as $R$4 (Jabi et al., 2011). For Nakagami-$R$5 channels, $R$6 (José~David~Vega-Sánchez et al., 2023, Jabi et al., 2011).

In block-fading or repetition-based schemes, repetition across $R$7 independent fading blocks directly yields $R$8 (Choi, 2020, Duyck et al., 2011).

For cooperative relaying and network coding over $R$9 relays, opportunistic selection or best-hop schemes can provide diversity order $\rho$0, as seen in DF/AF relaying with perfect (or sufficiently accurate) CSI estimation (Kalantari et al., 2015, Michalopoulos et al., 2011). In network-coded multiuser relaying over Nakagami-$\rho$1 fading, the diversity order scales as $\rho$2 for $\rho$3 source-destination pairs, independent of the number of relays (Benamira et al., 2024).

3. Impact of Protocols, Fading, and Interference on Outage and Diversity

The diversity order and the form of $\rho$4 are highly protocol- and environment-dependent:

• Hybrid ARQ (HARQ) and Time/Frequency Diversity: With $\rho$5 orthogonal time/frequency resources, incremental redundancy or Chase combining achieves diversity $\rho$6 until the code rate $\rho$7 crosses certain thresholds that quantize the mutual-information or SNR accumulation (Shi et al., 2022, Shi et al., 2022, 0710.1595). In time-correlated channels, full $\rho$8 is still achieved as long as the channel correlation coefficient $\rho$9; increased correlation degrades coding gain but not the diversity exponent (Shi et al., 2022).
• Keyhole and Rank-Deficient MIMO: In MIMO systems under keyhole effect, only $$P_{\mathrm{out}}(\rho,R) = \mathbb{P}\{ I(\text{channel};\rho) < R\}$$0 spatial degrees of freedom survive, even with HARQ. The diversity order in $$P_{\mathrm{out}}(\rho,R) = \mathbb{P}\{ I(\text{channel};\rho) < R\}$$1-round HARQ becomes $$P_{\mathrm{out}}(\rho,R) = \mathbb{P}\{ I(\text{channel};\rho) < R\}$$2, a stark contrast to the $$P_{\mathrm{out}}(\rho,R) = \mathbb{P}\{ I(\text{channel};\rho) < R\}$$3 attainable in non-degenerate channels (Zhang et al., 2022).
• NOMA and Cooperative Schemes: In repetition-based NOMA, each additional repetition strengthens diversity, with $$P_{\mathrm{out}}(\rho,R) = \mathbb{P}\{ I(\text{channel};\rho) < R\}$$4 and $$P_{\mathrm{out}}(\rho,R) = \mathbb{P}\{ I(\text{channel};\rho) < R\}$$5 (Choi, 2020). Power-domain NOMA with HARQ in downlink settings reveals that the diversity order is a step function of rate, and users with weaker mean channel gain ultimately limit the diversity of all stronger users (Shi et al., 2022).
• Interference-Limited and Spatially-Correlated Networks: In stochastic interference-limited settings (e.g., Poisson networks), the spatial-contention diversity order (SC-DO) replaces classical diversity. With Rayleigh fading and static interferer geometry, even $$P_{\mathrm{out}}(\rho,R) = \mathbb{P}\{ I(\text{channel};\rho) < R\}$$6 retransmissions cannot increase the diversity beyond the ‘interference exponent’ $$P_{\mathrm{out}}(\rho,R) = \mathbb{P}\{ I(\text{channel};\rho) < R\}$$7 (with path loss exponent $$P_{\mathrm{out}}(\rho,R) = \mathbb{P}\{ I(\text{channel};\rho) < R\}$$8), i.e., $$P_{\mathrm{out}}(\rho,R) = \mathbb{P}\{ I(\text{channel};\rho) < R\}$$9 (Haenggi et al., 2013, Tanbourgi et al., 2013). Diversity enhancement appears only when the spatial or temporal interference field is randomized (e.g., highly dynamic interferers or path-loss dominant/uncorrelated scenarios).
• Antenna Arrays, Spatial Correlation, and URLLC Regime: Formal classical diversity order $I(\cdot)$0 for an $I(\cdot)$1-branch array can misestimate actual performance in the ultra-reliability regime ($I(\cdot)$2). "Local diversity" at a finite threshold generalizes the slope of the CDF, showing that large-M arrays can see $I(\cdot)$3 at practical outages, especially with strong correlation or non-zero Rician K-factors (Abraham et al., 2021, José~David~Vega-Sánchez et al., 2023).
• Cooperative Relaying and Practical Limitations: In realistic dual-hop AF systems with multiple antennas, full diversity (d=N) is achieved only with variable-gain relaying and all antennas at the relay (1-N-1 topology). Fixed-gain schemes or suboptimal relay placement yield at most $I(\cdot)$4 irrespective of the number of antennas (Zhong et al., 2012, Zhu et al., 2014).

4. Role of Channel State Information and Estimation Quality

Diversity order is particularly sensitive to CSI imperfection. In relay selection under Nakagami-$I(\cdot)$5 fading, the decay speed of the correlation $I(\cdot)$6 matters critically:

• If $I(\cdot)$7 with $I(\cdot)$8 (i.e., estimation error vanishes at least as fast as the SNR inverse), full diversity $I(\cdot)$9 is retained for $d$0 relays. If $d$1, diversity collapses to $d$2 (Michalopoulos et al., 2011). Outdated or insufficiently frequent estimation leads to significant diversity loss—thus pilot allocation and prediction must be tuned appropriately to preserve high-SNR outage decay.

Imperfect channel estimation shifts outage curves right (SNR loss), but as long as estimation error decays “fast enough” with SNR, the asymptotic diversity order is unaffected (Kalantari et al., 2015, Michalopoulos et al., 2011).

5. Optimization and Trade-offs in System Design

Outage probability and diversity order directly inform system design, including code/precoder optimization, ARQ/HARQ configuration, and resource allocation:

• Precoding and Coding Gain: On block-fading channels, full diversity $d$3 is achieved if the coordinate projections of the symbol constellation meet certain cardinality conditions. Orthogonal precoders, constellation expansion, and mutual information maximization jointly minimize the outage threshold and maximize coding gain, closing the performance gap to the i.i.d. Gaussian-input capacity curve (Duyck et al., 2011).
• Blocklength, Repetition, and Rate Selection: In repetition- or block-based schemes, increasing $d$4 (blocks/repetitions) or lowering the code rate $d$5 (thus threshold $d$6) can maintain outage below stringent targets, directly trading reliability for spectral efficiency (Choi, 2020, 0710.1595).
• Resource Allocation in Multihop and Network Coding: In multiuser network-coded relaying, the cross-links often dominate performance; increasing the fading parameter $d$7 or the power budget on these links yields the highest marginal reduction in $d$8. The diversity order is dictated by the number of sources and the underlying Nakagami $d$9, not the number of relay nodes (Benamira et al., 2024).
• Outage Events Beyond Probability—Rate and Duration: Secondary statistics such as average outage rate (AOR) and outage duration (AOD) capture temporal clustering of outages in mobile scenarios. While the AOR decays with slope $$d = -\lim_{\rho\to\infty} \frac{\log P_{\mathrm{out}}(\rho)}{\log \rho}$$0 (like $$d = -\lim_{\rho\to\infty} \frac{\log P_{\mathrm{out}}(\rho)}{\log \rho}$$1), AOD decays with slope $$d = -\lim_{\rho\to\infty} \frac{\log P_{\mathrm{out}}(\rho)}{\log \rho}$$2, highlighting the distinct timescales associated with deep fade events and their persistence (Zlatanov et al., 2010).

6. Practical and Asymptotic Limitations

Classical asymptotic diversity analysis must be carefully interpreted in practical (non-asymptotic) regimes:

• In fixed-outage analysis (rate increases with SNR to maintain constant $$d = -\lim_{\rho\to\infty} \frac{\log P_{\mathrm{out}}(\rho)}{\log \rho}$$3), the affine approximation $$d = -\lim_{\rho\to\infty} \frac{\log P_{\mathrm{out}}(\rho)}{\log \rho}$$4 is appropriate in high SNR, and the diversity order under this scaling is $$d = -\lim_{\rho\to\infty} \frac{\log P_{\mathrm{out}}(\rho)}{\log \rho}$$5; only at fixed rate does the high-SNR slope recover the true diversity gain (0710.1595).
• For ARQ/HARQ, outage capacity advantages are pronounced at moderate SNR, but asymptotically (as SNR $$d = -\lim_{\rho\to\infty} \frac{\log P_{\mathrm{out}}(\rho)}{\log \rho}$$6) the diversity benefit vanishes relative to open-loop time/frequency diversity (0710.1595).
• In large antenna systems and the ultra-reliable regime ($$d = -\lim_{\rho\to\infty} \frac{\log P_{\mathrm{out}}(\rho)}{\log \rho}$$7), the asymptotic tail diversity order $$d = -\lim_{\rho\to\infty} \frac{\log P_{\mathrm{out}}(\rho)}{\log \rho}$$8 may significantly mischaracterize the real (local) reliability slope, mandating numerical or semi-analytic approaches (Abraham et al., 2021).

7. Summary Table of Outage Probability and Diversity Order for Key Models

System/Protocol	Outage Probability (High SNR)	Diversity Order $$d = -\lim_{\rho\to\infty} \frac{\log P_{\mathrm{out}}(\rho)}{\log \rho}$$9
$d$0-branch MRC (Rayleigh)	$d$1	$d$2
Block-fading, $d$3 blocks	$d$4	$d$5
NOMA, $d$6 repetitions	$d$7	$d$8
DF/AF opportunistic relaying, $d$9 relays	$P_{\mathrm{out}}(\rho)\sim C\rho^{-d}$0	$P_{\mathrm{out}}(\rho)\sim C\rho^{-d}$1
Dual-hop AF (relay with $P_{\mathrm{out}}(\rho)\sim C\rho^{-d}$2 antennas, var-gain)	$P_{\mathrm{out}}(\rho)\sim C\rho^{-d}$3 (only in 1-N-1, otherwise $P_{\mathrm{out}}(\rho)\sim C\rho^{-d}$4)	See text
HARQ, $P_{\mathrm{out}}(\rho)\sim C\rho^{-d}$5 rounds, full time-diversity	$P_{\mathrm{out}}(\rho)\sim C\rho^{-d}$6	$P_{\mathrm{out}}(\rho)\sim C\rho^{-d}$7
Keyhole MIMO (HARQ, $P_{\mathrm{out}}(\rho)\sim C\rho^{-d}$8 rounds)	$P_{\mathrm{out}}(\rho)\sim C\rho^{-d}$9	$\rho$0
Poisson interference-limited ($\rho$1 path loss)	$\rho$2	$\rho$3
Network coding ($\rho$4 S–D pairs, Nakagami-$\rho$5)	$\rho$6	$\rho$7

In all cases, outage probability and diversity order must be interpreted within the context of the system model, SNR regime, and the presence or absence of interference, cooperation, coding, and feedback. High-SNR exponents provide first-order guidance, but non-asymptotic analyses or local-slope (“local diversity”) methods are indispensable for ultra-reliable and finite-SNR operation. These principles unify the comparative evaluation of contemporary and emerging wireless transmission schemes on an information-theoretic and outage-centric basis.

---

References:

(Choi, 2020, Shi et al., 2022, Shi et al., 2022, Duyck et al., 2011, Jabi et al., 2011, José~David~Vega-Sánchez et al., 2023, Tanbourgi et al., 2013, Zhong et al., 2012, Kalantari et al., 2015, Benamira et al., 2024, Michalopoulos et al., 2011, Haenggi et al., 2013, 0710.1595, Abraham et al., 2021, Zhang et al., 2022, Ernest et al., 2017, Zlatanov et al., 2010, Zhu et al., 2014).