Single Distance Priority (SDP)

• Single Distance Priority (SDP) is a mechanism where one distance metric—such as hop count or Euclidean distance—is used to govern decision-making in routing, sparsification, and safety messaging.
• In WDM multicast routing, SDP selects candidate destinations and connectors based on shortest-path trees, achieving significant reductions in network diameter and average delay without increasing cost.
• In DSRC and multi-priority graph sparsification, SDP is employed either as a MAC-layer prioritization rule or as a key subroutine, ensuring near-optimal performance and effective resource allocation.

Single Distance Priority ("SDP", Editor's term) denotes a class of mechanisms in which a single distance metric governs prioritization, tie-breaking, or the invocation of a lower-level optimization primitive. The term is not used uniformly in the literature. In "Distance priority based multicast routing in WDM networks considering sparse light splitting," the operative phrase is "distance priority based" multicast routing; in "Multi-Priority Graph Sparsification," the closest equivalents are "1-priority solution," "single-priority subroutine," and "oracle computing the minimum-weight sparsification over $T$"; in "Prioritization of Basic Safety Message in DSRC Based on Distance to Danger," the mechanism is prioritization based solely on distance to a danger source (0908.1471, Ahmed et al., 2023, Kim et al., 2020).

1. Terminological scope and conceptual variants

The label SDP refers to different technical objects in different problem domains. In the WDM multicast-routing setting, it is a concrete selection rule: among Candidate Destinations, preference is given to smaller $d(s,d)$ in the source-rooted shortest-path tree, and among Candidate Connectors, preference is given to smaller $dist_T(s,c)$ in the current light-tree. In multi-priority graph sparsification, it is not itself the full multi-level algorithm, but rather the single-priority subroutine that is called at each rounded priority level. In DSRC, it is a MAC-layer prioritization rule in which Euclidean distance to a danger source determines a vehicle’s priority category and hence its backoff range.

Setting	Paper terminology	Role of SDP
WDM multicast routing	"distance priority based"	Single distance-based prioritization for Candidate Destinations and Candidate Connectors
Multi-priority graph sparsification	"1-priority solution", "single-priority subroutine"	Black-box subroutine solving one level’s distance or connectivity requirement
DSRC safety messaging	prioritization based on "distance to danger"	Distance determines priority category and backoff selection

This distribution of meanings suggests that SDP is best treated as a family resemblance among methods rather than as a single canonical algorithm. A common feature is that one distance quantity is privileged over richer multi-criteria scoring; the actual optimization object, however, varies sharply across optical multicast routing, graph sparsification, and vehicular MAC scheduling.

2. Distance-priority multicast routing in WDM networks

In sparse-splitting WDM multicast routing, the network is modeled as an undirected graph $G=(V,E)$, a multicast session $m$ is specified by a source $s\in V$ and a destination set $D\subseteq V\setminus\{s\}$, and light-trees must respect wavelength continuity and splitting constraints. Multicast-capable (MC) nodes can split an incoming optical signal to multiple outgoing ports, whereas multicast-incapable (MI) nodes have drop-and-continue capability but cannot serve as branching points in a light-tree. A light-tree $T=(V_T,E_T)$ is rooted at $s$, uses the same wavelength on all links in $E_T$, and obeys the no-MI-branching constraint. Distances are shortest-path hop counts, denoted $d(s,d)$0, computed by Dijkstra’s algorithm with unit weights. During construction, $d(s,d)$1 contains all MC nodes in $d(s,d)$2 and all leaf MI nodes in $d(s,d)$3, while $d(s,d)$4 contains non-leaf MI nodes that are forbidden to be traversed when attaching new destinations (0908.1471).

The key structural notion is the Shortest Constraint Path (SCP). A constraint path between $d(s,d)$5 and the current tree $d(s,d)$6 is any shortest path $d(s,d)$7 to a node $d(s,d)$8 that does not traverse $d(s,d)$9. The set of such paths is

$dist_T(s,c)$0

and the SCP is the minimum-length path in that set:

$dist_T(s,c)$1

At each iteration, Candidate Destinations are the unjoined destinations with minimum SCP length:

$dist_T(s,c)$2

Among these, the distance-priority rule selects

$dist_T(s,c)$3

where $dist_T(s,c)$4 is measured in the precomputed source-rooted shortest-path tree $dist_T(s,c)$5. For the chosen $dist_T(s,c)$6, Candidate Connectors are those $dist_T(s,c)$7 that realize an SCP to $dist_T(s,c)$8, and the selected connector is

$dist_T(s,c)$9

where $G=(V,E)$0 is the hop count along the unique path within the current tree. Ties may be broken arbitrarily; the paper notes that better final trees can arise if equal-priority Candidate Destinations are "well organized," but no mandatory secondary metric is imposed.

Operationally, the algorithm builds $G=(V,E)$1 once, initializes a tree rooted at $G=(V,E)$2, and then iteratively attaches one destination at a time by computing constrained shortest paths on $G=(V,E)$3 with $G=(V,E)$4 as multi-source origins. If no destination is attachable, the current tree is added to a forest and a new tree is started from $G=(V,E)$5. The construction updates $G=(V,E)$6 and $G=(V,E)$7 so that any MI node that becomes non-leaf is removed from the attachable set. The overall complexity is $G=(V,E)$8, with $G=(V,E)$9 costing $m$0 and each attachment step requiring a constrained shortest-path computation of the same asymptotic order; the paper states that this does not increase time complexity relative to Member-Only.

The paper evaluates source-rooted diameter and average delay as

$m$1

$m$2

Link stress is $m$3, where $m$4 is the number of lightpaths using edge $m$5, and Total Cost is $m$6 for a single tree or the sum over trees in the forest. On COST-239, the reported diameter reduction is up to about $m$7 and the average-delay reduction is up to about $m$8, with link stress and total cost always identical to Member-Only in all tested cases. On USA Longhaul, the paper reports diameter reduction up to $m$9 as $s\in V$0, average-delay reduction up to $s\in V$1 as $s\in V$2, and generally the same or slightly better link stress and total cost, with some cases of slightly worse link stress or total cost when the number of destinations is $s\in V$3 and the number of MC nodes is very small. The stated explanation is that choosing near-to-source destinations and near-to-source connectors shrinks the source-rooted height of the resulting light-tree without increasing SCP length.

3. Single-priority distance subroutines in multi-priority graph sparsification

In multi-priority graph sparsification, SDP corresponds to the single-priority subroutine assumed by the framework rather than to the final multi-priority solution itself. The graph is $s\in V$4 with nonnegative edge weights $s\in V$5, and induced distances are $s\in V$6. Each vertex $s\in V$7 has a priority $s\in V$8, with terminal sets $s\in V$9. A $D\subseteq V\setminus\{s\}$0-priority sparsification assigns each edge $D\subseteq V\setminus\{s\}$1 a rate $D\subseteq V\setminus\{s\}$2, inducing nested subgraphs $D\subseteq V\setminus\{s\}$3, and validity requires $D\subseteq V\setminus\{s\}$4 for each level-specific sparsifier family $D\subseteq V\setminus\{s\}$5. Under the linear rate-cost model, the cost of using edge $D\subseteq V\setminus\{s\}$6 at rate $D\subseteq V\setminus\{s\}$7 is $D\subseteq V\setminus\{s\}$8, and the total weight is $D\subseteq V\setminus\{s\}$9 (Ahmed et al., 2023).

The SDP object is the "oracle that can compute the minimum weight sparsification of $T=(V_T,E_T)$0 over $T=(V_T,E_T)$1" or a $T=(V_T,E_T)$2-approximation thereof. Depending on the problem class, this can be a single-priority subset spanner, a distance preserver, or a Steiner-tree solver. For distance-approximation problems, the subroutine input is $T=(V_T,E_T)$3, a priority-specific terminal set $T=(V_T,E_T)$4 or a pair set $T=(V_T,E_T)$5, and the desired stretch parameters $T=(V_T,E_T)$6, and the output is a subgraph $T=(V_T,E_T)$7 such that

$T=(V_T,E_T)$8

for all required pairs. For exact distance preservers, $T=(V_T,E_T)$9 and $s$0. For single-priority Steiner trees, the output is a tree connecting the specified terminals.

The full multi-priority method rounds vertex priorities up to powers of two, producing $s$1 rounded levels. It then constructs level instances either by inclusive partitioning, where $s$2, or by pairwise partitioning, where each pair $s$3 is assigned to the level $s$4. For each rounded level $s$5, it calls the SDP subroutine to obtain a single-priority solution $s$6, and then merges solutions from top to bottom by assigning rate $s$7 to edges first introduced at level $s$8 and propagating them to lower levels. If the object must be a tree, cycles are pruned by deleting lower-rate edges while preserving the required higher-level connectivity.

The paper’s main theorem states that, under linear edge-rate costs and with an exact oracle, at most $s$9 oracle calls suffice to compute a $E_T$0-priority sparsification with weight at most

$E_T$1

If the oracle is a $E_T$2-approximation, the guarantee becomes $E_T$3. This bound is independent of $E_T$4. The validity lemmas establish that both inclusive partitioning and pairwise partitioning yield correct solutions: inclusive partitioning preserves cross-priority distance constraints because higher-priority vertices remain present at lower levels, while pairwise partitioning enforces each pair at its minimum-priority level.

The framework is instantiated in the paper for multiplicative subset spanners, additive subset spanners, near-additive subset spanners, weighted additive spanners using local-error formulations such as $E_T$5 and $E_T$6, distance preservers, Steiner trees, and $E_T$7-connected subgraphs. The per-level stretch delivered by the multi-priority construction matches the stretch of the underlying single-priority subroutine; the additional cost of priorities appears in the global $E_T$8 weight bound and the $E_T$9 number of SDP calls rather than in degraded per-level stretch. Emulators are mentioned as a classical sparsification object, but the formal guarantees in the paper apply to subgraphs and tree constraints rather than to non-subgraph shortcuts.

4. Distance-to-danger prioritization in DSRC

In DSRC safety messaging, SDP is a MAC-layer prioritization rule based solely on a vehicle’s distance to a danger source. The danger source is assumed to be at a known fixed location, placed at the origin of a two-dimensional Cartesian plane in the analysis. Vehicles know their own positions from the GPS fields in Basic Safety Messages, and the distance metric is the Euclidean norm

$d(s,d)$00

With latitude and longitude, the paper’s discussion permits either conversion to a local Cartesian map projection followed by Euclidean distance or use of a haversine or ellipsoidal distance to obtain the scalar $d(s,d)$01, though the analysis itself uses a planar Euclidean model (Kim et al., 2020).

Priority assignment is monotonic and threshold-based. With thresholds $d(s,d)$02 meters, the paper defines three categories: CAT 1 for $d(s,d)$03, CAT 2 for $d(s,d)$04, and CAT 3 for $d(s,d)$05. Vehicles with $d(s,d)$06 are non-prioritized and do not receive a special priority allocation. As an ordinal function,

$d(s,d)$07

Smaller $d(s,d)$08 therefore implies higher priority.

The implementation changes only backoff-counter selection in IEEE 802.11p/DSRC EDCA. Let $d(s,d)$09 denote the configured contention-window size for the access category used to send BSMs. The integer interval $d(s,d)$10 is partitioned into three equal contiguous subranges with boundaries

$d(s,d)$11

Then CAT 1 uses a uniform draw on $d(s,d)$12, CAT 2 on $d(s,d)$13, and CAT 3 on $d(s,d)$14. This makes nearer vehicles statistically choose shorter backoffs. The paper does not change AIFS, TXOP, or $d(s,d)$15 per access category; BSMs are broadcast every $d(s,d)$16 ms, there are no ACKs or retries for broadcast, and if a packet is not transmitted by the end of the current beaconing period, it expires and is dropped.

The analysis uses a single $d(s,d)$17 MHz channel, slot time $d(s,d)$18, $d(s,d)$19, $d(s,d)$20, propagation delay $d(s,d)$21, and payload length $d(s,d)$22 bytes. Vehicle positions form a homogeneous PPP with density $d(s,d)$23 in a $d(s,d)$24 km $d(s,d)$25 $d(s,d)$26 km region, corresponding to approximately $d(s,d)$27 vehicles. The proposed analysis treats collisions as negligible compared to expiration, that is, $d(s,d)$28. Average delay is written as

$d(s,d)$29

with

$d(s,d)$30

Normalized throughput is

$d(s,d)$31

and the inter-reception-time distribution in beacon periods is

$d(s,d)$32

with $d(s,d)$33.

The reported trends are that high-priority vehicles, especially CAT 1, experience smaller expected backoff, lower average delay, higher transmission probability $d(s,d)$34, improved normalized throughput, and shorter inter-reception time than under traditional uniform backoff. With $d(s,d)$35, $d(s,d)$36 for CAT 1 is consistently higher under the proposed scheme than under the traditional scheme across the number of contending stations. With $d(s,d)$37, the IRT probability mass function for CAT 1 places most mass at $d(s,d)$38–$d(s,d)$39 beacon periods under the proposed scheme. The paper also states that the advantage increases under higher contention, while fairness is deliberately shifted toward nearer vehicles.

5. Structural commonalities and domain-specific distinctions

Across these literatures, the operative feature of SDP is not a shared data structure or protocol stack, but the use of one distance quantity as the privileged control signal. In sparse-splitting WDM multicast routing, the relevant quantities are $d(s,d)$40 in the precomputed shortest-path tree and $d(s,d)$41 in the current light-tree. In multi-priority graph sparsification, the single-priority subroutine is distance-aware through its spanner or preserver guarantees on a given terminal set or pair set. In DSRC, the priority variable is $d(s,d)$42, which directly determines contention behavior (0908.1471, Ahmed et al., 2023, Kim et al., 2020).

This suggests a useful but narrow unifying interpretation. SDP does not mean that the entire optimization is one-dimensional; rather, a single distance metric controls a critical decision layer. In the WDM case, the metric breaks ties among shortest feasible attachments without altering the underlying Minimum Path heuristic. In graph sparsification, the metric appears inside a black-box level solver whose output is then lifted to a multi-priority construction by rounding and merging. In DSRC, the metric does not solve a routing problem at all; it reshapes random access by restricting the support of backoff draws.

A related misconception is that SDP necessarily denotes a globally optimal delay-minimization procedure. The cited works do not support that interpretation. The WDM algorithm preserves link stress and wavelength usage "inherited from the Minimum Path heuristic" while improving source-rooted diameter and average delay; the graph-sparsification framework gives a $d(s,d)$43-approximation under a linear rate-cost model rather than exact multi-priority optimality; and the DSRC method redistributes transmission opportunities toward dangerous vehicles without increasing overall load, but under an expiration-constrained analytical regime.

6. Assumptions, limitations, and research implications

Each realization of SDP depends on restrictive modeling assumptions. The WDM multicast-routing formulation assumes sparse light splitting, no wavelength conversion, wavelength continuity per light-tree, and hop count as the delay and cost metric. Its benefits depend on topology, multicast-group size, and splitter distribution; when the multicast group is small, there may be few Candidate Destinations or Candidate Connectors, limiting the effect of the priority mechanism, and in some cases with very few MC nodes and many destinations, link stress or total cost can be slightly worse than Member-Only (0908.1471).

The graph-sparsification framework assumes undirected graphs and the linear rate-cost model $d(s,d)$44 throughout its approximation analysis. The rounding theorem’s $d(s,d)$45 guarantee is proved under that model, and emulators are noted but not formally included in the subgraph-based framework. Directed graphs are not addressed, and some weighted additive results rely on local-error formulations using $d(s,d)$46 rather than plain additive constants (Ahmed et al., 2023).

The DSRC formulation assumes a single danger source, fixed $d(s,d)$47 Hz beaconing, no ACK or retry for broadcast, and an expiration-dominated regime in which $d(s,d)$48. The paper does not model detailed PHY capture, fading, or hidden-node collisions beyond that approximation, and it does not analyze multiple concurrent dangers. It also explicitly shifts access in favor of near-danger vehicles, which can increase delay and inter-reception time for far vehicles. The implementation cost is $d(s,d)$49 per beacon period per vehicle, but the safety benefit depends on correct danger-source dissemination and position integrity (Kim et al., 2020).

A plausible implication is that SDP is best understood as a design pattern for priority formation rather than as a domain-independent theory. The cited works show that a single distance metric can be operationally sufficient for improving source-rooted delay metrics in WDM multicast routing, for serving as the primitive solver inside a multi-priority sparsification meta-algorithm, and for biasing channel access toward vehicles nearest to a hazard. They do not show that single-distance prioritization is universally superior to richer multi-criteria schemes; rather, they delineate settings in which the reduction to one distance quantity is analytically tractable and empirically useful.