PCP-TDMA Protocol for Distributed Wireless Scheduling

• PCP-TDMA is a distributed MAC protocol that uses local one- and two-hop neighbor information to establish collision-free link activations in wireless networks.
• It employs iterative slot reservation and superframe truncation phases to compact schedules and maximize concurrent transmissions, especially in MTR mesh networks.
• Empirical studies show that PCP-TDMA can improve throughput by up to 110% over traditional protocols while offering scalability and adaptability in dynamic topologies.

PCP-TDMA (Pseudo TDMA with Peer-to-Peer Control) refers to a family of fully distributed medium access control protocols designed to establish deterministic, collision-free schedules for link activation in wireless multi-hop networks. Its key innovation lies in leveraging only local (one-hop and limited two-hop) neighbor information and peer-to-peer coordination, enabling network nodes to agree on TDMA-like schedules without centralized control or global topology knowledge. PCP-TDMA paradigms have been developed for both standard wireless mesh topologies and for emerging Multi-Transmit-Receive (MTR) mesh routers with multiple directional antennas, optimizing network capacity and schedule compactness while providing adaptivity and scalability (Xu et al., 2016, Hui et al., 2011).

1. Network Model and Assumptions

PCP-TDMA protocols are defined on a network model where nodes are represented as vertices $V$ of a graph $G=(V,E)$ with directed links $E$, and connectivity is governed by a distance/range constraint ($(u,v)\in E$ if $\mathrm{dist}(u,v)\leq r$). Each node $u$ has degree

$$d_u = |N(u)|,\quad N(u)=\{v\in V:(u,v)\in E\text{ or }(v,u)\in E\},$$

and is typically equipped with $b_u \geq d_u$ radios and directional antennas, supporting simultaneous transmissions to or from multiple neighbors. However, the half-duplex (no Mix-Tx-Rx) constraint remains: a node may not simultaneously transmit and receive, even if using different antennas.

Time is discretized into synchronized slots, forming a periodic superframe of length $P$, where

$$\mathcal{SF} = \{\epsilon_1,\,\epsilon_2,\,\dots,\,\epsilon_P\}$$

and each $G=(V,E)$0 is a maximal feasible set of link activations in slot $G=(V,E)$1 under all interference and half-duplex constraints. Each node tracks its sets of transmitting and receiving slot assignments as

$G=(V,E)$2

A directed link $G=(V,E)$3 can only be scheduled in slots not present in $G=(V,E)$4 to preclude collisions and duplex conflicts.

2. Design Principles and Objectives

PCP-TDMA protocols are structured around three central aims:

1. Exploitation of MTR: Maximizing the set of concurrent, non-conflicting link activations permitted by the physical layer, especially when nodes possess multiple directional antennas.
2. Collision Avoidance: Ensuring no two links violate interference constraints or cause a node to transmit and receive simultaneously.
3. Superframe Minimization: Reducing the superframe length $G=(V,E)$5 so every feasible link is activated at least once in as few slots as possible.

Unlike classical, often centralized, TDMA and contention-based (e.g., CSMA) MAC protocols, PCP-TDMA systems are fully distributed. Nodes perform iterative, local allocation of transmission slots using only neighbor information, and periodically consolidate slot reservations by attempting to move existing activations toward the front of the frame. Once no link can advance its slot further, nodes agree—by distributed consensus—to truncate (shorten) the superframe, optimizing schedule compactness (Xu et al., 2016, Hui et al., 2011).

3. Distributed Scheduling Algorithms

The canonical PCP-TDMA protocol for MTR mesh networks (as in (Xu et al., 2016)) involves a two-phase distributed scheduling mechanism:

3.1 Slot Reservation Phase

Each unscheduled directed link $G=(V,E)$6 is assigned, possibly randomly, to a slot later in the superframe. In every superframe iteration, node $G=(V,E)$7 attempts to move $G=(V,E)$8 to an earlier feasible slot by evaluating

$G=(V,E)$9

and selecting a slot $E$0 with $E$1 (earlier in the superframe), where all constraints remain satisfied.

Nodes communicate slot-reservation intent to their one-hop neighbors via local control messages, resolving conflicts as necessary. This reservation-and-shift process is repeated over successive superframes until no further progress is possible; that is, all links occupy locally earliest feasible slots.

3.2 Superframe Truncation Phase

Once all link activations are as close as possible to the beginning of the superframe, all nodes jointly agree to reduce the superframe length $E$2 by eliminating trailing empty slots. This synchronization is achieved using distributed messages confirming stability and slot assignment completeness across all nodes.

3.3 Peer-to-Peer State Exchange and Voting (Alternate PCP-TDMA)

A related family of protocols employs a peer-to-peer, vote-based state exchange, in which each node $E$3 exchanges its current state (slot assignment) $E$4 with all one- and two-hop peers. Nodes select their next slot in frame $E$5 by aggregating votes from neighbors, favoring slots that are idle or uniquely owned. In the broadcast setting, the natural distributed rule for determining the slot resolution is

$E$6

so each node’s slot space adapts to local density (Hui et al., 2011).

4. Performance and Theoretical Guarantees

PCP-TDMA achieves, in a fully connected network, superframe lengths less than one third and one half those produced by distributed competitors JazzyMAC and ROMA, respectively. In arbitrary topologies, it demonstrates near-equivalence to centralized, globally optimized schedules while remaining fully distributed (Xu et al., 2016).

The convergence process is characterized by Markovian evolution in the global slot-state space, with formal proofs that the protocol almost surely attains absorbing, collision-free states (schedules) under mild assumptions:

• 1D Topologies: The multi-resolution state-exchange variant converges with probability one if per-node slot resolution $E$7 meets the logarithmic lower bound.
• 2D Topologies: Provided slot resolutions are sufficiently large, and stochastic tie-breaking is permitted, all absorbing states are collision-free (Hui et al., 2011).

Throughput is quantified as

$E$8

for broadcast, where the angle brackets denote spatial averaging over all nodes.

Empirical results indicate that MTR-based PCP-TDMA substantially increases capacity relative to conventional distributed MACs, and in 1D/2D random networks, achieves convergence and collision-free operation efficiently. For saturated broadcast with moderate node densities, throughput improvements range from 50% to 110% over optimized slotted ALOHA (Hui et al., 2011).

5. Protocol Overhead, Scalability, and Practical Considerations

Each node transmits $E$9 bits of control overhead per frame, where the cost grows slowly with local density and the logarithm of neighbor count. All peer state exchanges are strictly local and require only two-hop topology knowledge, enabling high scalability without centralized scheduling.

Computational load per scheduling round is linear in $(u,v)\in E$0, dominated by collision-checking for candidate slots. Adaptivity is inherent: upon local topology changes, nodes dynamically adjust their slot resolutions and repeat the local voting and reservation process.

A known limitation is that the search for globally optimal throughput under general interference models remains NP-hard; PCP-TDMA is guaranteed only to find feasible, collision-free schedules (Hui et al., 2011).

6. Variants and Relation to Other MAC Protocols

PCP-TDMA builds on and extends ideas from earlier peer-to-peer state exchange TDMA (sometimes also referred to as “PCP-TDMA” in the literature), but is uniquely tailored for MTR networks with multiple directional radios (Xu et al., 2016). It generalizes the classical TDMA framing and assignment paradigm by introducing fine-grained local slot allocation and peer-coordination heuristics, distinct from contention-based (CSMA), fixed TDMA, and heuristic link scheduling frameworks.

The protocol outperforms distributed algorithms such as JazzyMAC and ROMA by more compactly packing link activations into minimum-length superframes and achieving near-centralized throughput. The fully distributed, self-organizing nature and robustness to topology changes set PCP-TDMA apart for practical, large-scale wireless mesh and sensor network deployment.

7. Analytical and Empirical Outcomes

Formal analysis confirms that PCP-TDMA protocols robustly converge to collision-free schedules in arbitrary static or slowly time-varying topologies. Simulation studies demonstrate high capacity, rapid convergence—including with annealing-based randomization for escaping local minima—and strong adaptivity to changes in local node density and traffic demands (Hui et al., 2011).

In summary, PCP-TDMA represents a rigorous approach to distributed channel access and link scheduling in both conventional and MTR-enabled wireless mesh networks, providing near-optimal superframe compactness, local adaptability, peer-driven coordination, and scalability without centralized control (Xu et al., 2016, Hui et al., 2011).