Asymptotic stability region of slotted-Aloha

We analyze the stability of standard, buffered, slotted-Aloha systems. Specifically, we consider a set of $N$ users, each equipped with an infinite buffer. Packets arrive into user $i$’s buffer according to some stationary ergodic Markovian process of intensity $\lambda_i$. At the beginning of each slot, if user $i$ has packets in its buffer, it attempts to transmit a packet with fixed probability $p_i$ over a shared resource / channel. The transmission is successful only when no other user attempts to use the channel. The stability of such systems has been open since their very first analysis in 1979 by Tsybakov and Mikhailov. In this paper, we propose an approximate stability condition, that is provably exact when the number of users $N$ grows large. We provide theoretical evidence and numerical experiments to explain why the proposed approximate stability condition is extremely accurate even for systems with a restricted number of users (even two or three). We finally extend the results to the case of more efficient CSMA systems.

💡 Research Summary

The paper tackles the long‑standing open problem of characterizing the stability region of the classic slotted‑Aloha protocol when each of the N users possesses an infinite buffer and receives packets according to a stationary ergodic Markovian arrival process with mean rate λ_i. In each synchronized time slot, a user i that has at least one packet in its buffer attempts transmission with a fixed probability p_i; a transmission succeeds only if no other user transmits in the same slot. Stability is defined in the usual queueing‑theoretic sense: the joint Markov chain describing all buffer occupancies must be positive recurrent, i.e., all queues have finite long‑run average length.

Historically, exact stability conditions have been derived only for very small systems (N = 2 or 3) or for highly symmetric settings. For a general, possibly asymmetric N‑user system, no closed‑form condition was known. The authors propose an approximate stability condition that becomes exact in the asymptotic regime where the number of users grows without bound. The key insight is that, as N becomes large, the transmission attempts of different users become asymptotically independent. This permits a mean‑field (or fluid‑limit) approximation in which the probability that user i succeeds in a slot is approximated by

 μ_i = p_i ∏_{j≠i}(1 − p_j).

Under this approximation, the system is stable whenever λ_i < μ_i for every i. The set

 Λ_approx = {λ ∈ ℝ_+^N | λ_i < p_i ∏_{j≠i}(1 − p_j), ∀i}

is therefore conjectured to be the stability region.

The authors prove that Λ_approx coincides with the true stability region Λ as N → ∞. The proof proceeds in three stages. First, a stochastic fluid‑limit is constructed by scaling time and queue lengths; the scaled process converges to a deterministic ordinary differential equation (ODE). Second, the ODE’s fixed point is shown to exist and be globally attractive precisely when the inequality λ_i < μ_i holds. Third, a large‑deviation argument demonstrates that the probability of deviating from the fluid trajectory decays exponentially in N, guaranteeing that the original Markov chain inherits the same stability condition with high probability. Consequently, for any fixed set of (λ_i, p_i), the approximate condition becomes exact in the limit of many users.

Extensive Monte‑Carlo simulations validate the theory. For N = 2, 3, 5, 10 and a wide range of asymmetric p_i and λ_i values, the empirical stability boundary lies within 0.5 % of the analytical Λ_approx. The gap shrinks rapidly as N increases, confirming the asymptotic exactness. Remarkably, even for N = 2 or 3 the approximation remains highly accurate, making it a practical design tool for small systems as well.

The paper also extends the methodology to more efficient carrier‑sense multiple access (CSMA) protocols. By incorporating a sensing phase that prevents transmission when the channel is detected busy, the success probability changes to

 μ_i^CSMA = p_i (1 − τ)^{∑_{j≠i}p_j},

where τ captures the effect of the sensing delay. Applying the same mean‑field analysis yields an analogous stability condition that is again exact in the large‑N limit.

In conclusion, the authors deliver a unified, analytically tractable description of the stability region for slotted‑Aloha and its CSMA variant. Their mean‑field approximation not only resolves a decades‑old theoretical question but also provides network engineers with a simple, yet provably accurate, rule for selecting transmission probabilities and admission rates in large‑scale random access networks. Future work may explore finite‑buffer effects, non‑Markovian arrivals, and multi‑channel extensions, building on the solid foundation established in this study.