PaperlayerView exact revision on arXiv

Source-linked AI summary

Multi-Hop Cluster based IEEE 802.11p and LTE Hybrid Architecture for VANET Safety Message Dissemination

Seyhan Ucar, Sinem Coleri Ergen, Oznur Ozkasap

arXiv:1603.05628v1cs.NI

TL;DR

VANET safety dissemination must remain timely and reliable despite fragmented, rapidly changing networks and the limitations of IEEE 802.11p alone. The paper introduces VMaSC-LTE, combining multi-hop IEEE 802.11p clustering with LTE through dual-interface cluster heads, and evaluates it against alternative architectures and routing mechanisms. Simulations report superior performance, while showing that higher delivery reliability requires greater LTE usage.

  • Problem

    VANET safety applications face broadcast storms, disconnected networks, and strict delay and packet-delivery requirements.

  • Method

    VMaSC-LTE combines IEEE 802.11p multi-hop clustering with LTE, using VMaSC to select and maintain clusters and dual-interface cluster heads to connect VANET and LTE.

  • Results

    Extensive ns-3 simulations with SUMO mobility show superior VMaSC-LTE performance over prior hybrid architectures, flooding, and cluster-based routing in delivery ratio, delay, overhead, and stability.

  • Takeaways & Limitations

    VMaSC-LTE can adapt reliability through cluster sizing, but improving delivery ratio increases LTE usage as the number of cluster heads rises.

Abstract

from arXiv · show

This paper proposes a hybrid architecture, namely VMaSC-LTE, combining IEEE 802.11p based multi-hop clustering and the fourth generation cellular system, Long Term Evolution (LTE), with the goal of achieving high data packet delivery ratio and low delay while keeping the usage of the cellular architecture at the minimum level. In VMaSC-LTE, vehicles are clustered based on a novel approach named VMaSC: Vehicular Multi-hop algorithm for Stable Clustering. From the clustered topology, elected cluster heads operate as dual-interface nodes with the functionality of IEEE 802.11p and LTE interface to link VANET to LTE network. Using various key metrics of interest including data packet delivery ratio, delay, control overhead and clustering stability, we demonstrate superior performance of the proposed architecture compared to both previously proposed hybrid architectures and alternative routing mechanisms including flooding and cluster based routing via extensive simulations in ns-3 with the vehicle mobility input from the Simulation of Urban Mobility (SUMO). The proposed architecture also allows achieving higher required reliability of the application quantified by the data packet delivery ratio at the cost of higher LTE usage determined by the number of cluster heads in the network.

I. INTRODUCTION

VANET safety dissemination must meet strict delay and packet-delivery requirements despite broadcast storms, disconnected networks, and rapidly changing topology. The paper proposes VMaSC-LTE, a hybrid IEEE 802.11p–LTE architecture designed to improve reliability and delay while limiting cellular usage.

  • Motivation: VANET safety applications require timely dissemination despite rapid topology changes, uneven density, and frequent network fragmentation.Safety-message packet delivery ratios typically range from 90% to 100%, while delay tolerance depends on the application.
  • Motivation: IEEE 802.11p dissemination faces broadcast storms at high density and disconnected networks at low density.Existing responses to these problems can produce delays from seconds to minutes and delivery rates as low as 60%.
  • Research gap: Prior hybrid architectures combine IEEE 802.11p and cellular communication, but do not jointly analyze clustering stability, delay, cluster-head effects, and message dissemination.The paper identifies missing stability analysis and limited treatment of multi-hop clustering effects on cluster-head count and stability.
  • Proposed approach: VMaSC selects cluster heads using relative mobility and reduces overhead through direct cluster connections, periodic member information, reactive maintenance, and size- and hop-limited merging.These mechanisms are combined in the proposed multi-hop IEEE 802.11p–LTE hybrid architecture.
  • Evaluation: The study evaluates delivery ratio, delay, control overhead, and clustering stability against prior hybrid architectures, flooding, and cluster-based routing using ns-3 with SUMO mobility.The evaluation uses a large-scale highway and a realistic vehicle mobility model.
  • Design trade-off: The architecture exposes a reliability–LTE-usage trade-off: higher delivery reliability requires more cluster heads and therefore greater cellular usage.The paper measures LTE cost by the number of cluster heads in the network.

II. SYSTEM MODEL

The system forms multi-hop clusters over IEEE 802.11p, while LTE-connected cluster heads disseminate safety data geographically. VMaSC targets high delivery and low delay while reducing cluster-head and infrastructure overhead.

  • Architecture: Vehicles form multi-hop clustered topologies, with nearby vehicles as cluster members and farther vehicles relaying through multi-hop members to reach a cluster head.Cluster members communicate directly with their cluster head; multi-hop members forward packets through connected members.
  • Interfaces: Cluster heads use both IEEE 802.11p and LTE, whereas cluster members communicate within their clusters through IEEE 802.11p.Cluster heads connect the vehicular network to the LTE network through eNodeBs.
  • LTE dissemination: LTE disseminates packets through eNodeBs and the EPC to eNodeBs covering the safety dissemination region.The EPC determines the relevant eNodeBs, which then disseminate the packet within the target region.
  • Design objective: VMaSC aims for high packet delivery and low delay while minimizing cluster heads and maximizing clustering stability.These objectives are intended to reduce overhead on vehicles and eNodeBs.
  • VMaSC: VMaSC provides stable clustering through relative mobility, low-overhead cluster connections, and reactive maintenance.The algorithm is introduced as the vehicular multi-hop algorithm for stable clustering.

B. V IB Generation and Update

VMaSC builds and refreshes each vehicle’s multi-hop neighborhood database, then uses relative speed and state information to select and maintain cluster roles as topology changes.

  • VIB generation: The VIB stores vehicle and neighbor information within MAX HOP hops, including mobility, clustering state, parent, and packet information.This repository supports determining cluster members and heads.
  • VIB update: VIB entries are updated by local information changes and periodic HELLO PACKET reception, then removed after VIB TIMER expiration.HELLO PACKET messages are retransmitted to neighbors within MAX HOP hops.
  • Mobility metric: The clustering metric AV GREL SPEED_i is the average relative speed of vehicle i compared with same-direction neighbors within MAX HOP hops.Lower average relative speed indicates lower mobility relative to neighboring vehicles.
  • Head selection: The vehicle with the lowest average relative speed is elected cluster head.The metric favors a vehicle that is less mobile relative to its neighbors.
  • State transitions: Vehicles transition from initial state to state election after building the VIB through periodic HELLO PACKET exchange.State election uses the gathered information to choose the next state.
  • State transitions: A vehicle becomes a cluster member after a successful JOIN RESP or becomes a cluster head when the head condition is satisfied.The head condition includes inability to connect to existing clusters, an available election-state neighbor, and minimum average relative speed.
  • Cluster maintenance: An isolated cluster head can return to state election after CH TIMER without members, reducing the number of clusters by connecting elsewhere.ISO-CH can become a cluster head after gaining a member or return to election when it is no longer isolated.
  • Cluster maintenance: A cluster member returns to state election after losing its parent connection for CM TIMER.The parent connection is considered lost when no packet is received during that interval.

D. Cluster Formation

Cluster formation first reuses existing clusters, prioritizing nearby heads and then multi-hop member paths; otherwise, vehicles elect heads using relative mobility or remain in election.

  • Cluster Formation: A vehicle in state election first attempts to join an existing cluster to minimize the number of cluster heads.Neighboring cluster heads receive priority over cluster members to reduce packet-forwarding hops and delay.
  • Cluster Formation: Vehicles scan neighboring cluster heads by increasing average relative mobility and join an eligible head after receiving JOIN RESP.Eligibility depends on member capacity and whether the vehicle has already tried that head.
  • Cluster Formation: If direct head connection fails, a vehicle connects to a cluster head through a cluster member using multiple hops.The multi-hop attempt also checks member capacity and hop limits.
  • Cluster Formation: When no head or member is reachable, a vehicle becomes an isolated cluster head if it has the smallest average relative speed among neighboring election-state vehicles.Otherwise, it remains in state election for SE TIMER and reruns the algorithm.

E. Cluster Merging

VMaSC tracks cluster membership centrally at each head and merges neighboring clusters when stability and cluster-count trade-offs permit, reducing spatial overlap and interference.

  • Cluster information: A cluster head stores member IDs, parent nodes, and cluster direction in CLUSTER INFO.Because members do not send JOIN REQ messages to the head through multiple hops, HELLO PACKET messages provide cluster information.
  • Merge decision: Neighboring cluster heads wait MERGE TIMER before exchanging CLUSTER INFO and average relative speed for possible merging.A longer MERGE TIMER increases cluster stability but also increases the number of clusters.
  • Inter-cluster interference: VMaSC reduces overlapping clusters by having election-state vehicles join existing clusters first and allowing feasible neighboring heads to merge.Reducing overlap addresses higher medium contention and inefficient flooding caused by inter-cluster interference.

G. Theoretical Analysis of VMaSC Clustering

The analysis models communication-link stability between neighboring vehicles and shows that minimizing relative speed maximizes the probability of maintaining connectivity.

  • The analysis models neighboring vehicles with speeds v1 and v2, accelerations a1 and a2, and separation r12 constrained to the transmission range.Vehicle separation is treated as symmetric around zero, while speeds are predetermined and accelerations are random variables.
  • At time T, connectivity requires the vehicle separation expression r12 + (v1 −v2)T + (a1 −a2)T 2/2 to remain within the transmission range.The communication probability is formulated as P(−rt < r12 + (v1 −v2)T + (a1 −a2)T 2/2 < rt).
  • When accelerations are equal, maximizing connection probability requires minimizing the relative speed |v1 −v2|.
  • When accelerations differ, the probability bounds depend on the relative-speed and relative-acceleration terms over time.The bounds use lower and upper bounds involving |(v1 −v2)T| and |δ12T 2/2|.
  • The resulting design criterion is to minimize relative speed, which maximizes both the lower and upper bounds of the connectivity probability.

IV. DATA DISSEMINATION IN HYBRID ARCHITECTURE

The hybrid architecture forwards safety data through clustered IEEE 802.11p links and LTE infrastructure, using cluster heads to bridge vehicles, eNodeBs, and the EPC.

  • LTE supplements IEEE 802.11p by maintaining connectivity when the VANET is disconnected and improving transmission delay and reliability.
  • Data from a cluster member is unicast to its cluster head, broadcast within the cluster, and forwarded through the eNodeB and EPC.
  • Cluster members and heads check packet identifiers and sequence numbers before forwarding, while heads create LTE packets for infrastructure delivery.
  • The EPC multicasts received LTE data to neighboring eNodeBs covering the packet’s target geographical area.
  • Destination eNodeBs multicast packets to their covered cluster heads, which broadcast them to cluster members.

V. PERFORMANCE EVALUATION

The evaluation uses ns-3 with SUMO-generated vehicle mobility to compare VMaSC-LTE against clustering-based hybrids, VANET routing, and flooding.

  • The simulations use ns-3 Release 3.17 and realistic vehicle mobility generated by the SUMO traffic simulator.SUMO models individual-driver behavior, including acceleration and overtaking decisions.
  • The study compares VMaSC-LTE with NHop, MDMAC, their hybrid architectures, flooding, and pure VANET cluster-based forwarding.
  • The scenario uses a 5 km, two-lane, two-way road with vehicles injected at 2 vehicles per second and 300 seconds of metric evaluation.
  • The evaluation examines clustering stability, delay, and data packet delivery ratio across the proposed and comparison mechanisms.

A. VANET Clustering

VMaSC is evaluated against NHop and MDMAC using cluster-duration, change-rate, overhead, and isolated-vehicle metrics; its cluster heads last longer under all tested conditions.

  • The comparison measures cluster head duration, cluster member duration, cluster head change rate, clustering overhead, and vehicles in SE state.
  • Cluster head duration is the interval from entering CH state until transitioning to SE or CM state.
  • VMaSC has higher average cluster head duration than NHop and MDMAC under all tested maximum velocities and hop limits.The paper attributes this primarily to its efficient cluster maintenance mechanism.
  • Cluster member duration measures the interval from joining an existing cluster in CM state until leaving it for SE state.

2) Cluster Member Duration:

VMaSC provides more stable clustering than NHop and MDMAC across mobility and hop-count conditions, while reducing clustering overhead through efficient maintenance and connections.

  • Cluster Member Duration: VMaSC achieves higher average cluster member duration than NHop and MDMAC under all evaluated conditions.The comparison varies maximum vehicle velocity and maximum allowed hops.
  • Cluster Head Change Rate: VMaSC has a lower cluster head change rate than NHop and MDMAC in all cases, indicating higher clustering stability.Cluster head change rate counts CH-to-other-state transitions per unit time.
  • Clustering Overhead: VMaSC produces smaller clustering overhead than NHop and MDMAC.Its advantage is attributed to greater cluster stability, efficient neighboring-member connections, and timer-based maintenance without periodic active clustering.
  • Clustering Overhead: Clustering overhead increases with maximum hop count because HELLO packets are rebroadcast over multiple hops.The overhead of NHop and MDMAC increases more steeply with vehicle velocity than VMaSC’s overhead.
  • SE-State Vehicles: At higher vehicle velocities, VMaSC has more vehicles in SE state when the maximum hop count is small, while this difference and temporal variation decrease as hop count increases.More hops provide a larger neighbor set for cluster selection and more potential cluster members.

1) Data Packet Delivery Ratio (DPDR):

VMaSC-LTE delivers the highest DPDR across evaluated velocities, hop counts, and vehicle densities while combining high delivery with low delay through LTE inter-cluster communication.

  • Hybrid Comparison: VMaSC-LTE’s superior DPDR over CMGM-LTE and MDMAC-LTE is attributed to better clustering stability, minimal overhead, and minimal cluster overlap.These properties reduce SE-state nodes and medium-access contention, increasing transmission success probability.
  • Vehicle Density: VMaSC performs poorly at low and high vehicle densities because of disconnected-network and broadcast-storm problems, respectively.VMaSC-LTE instead provides high DPDR that remains stable across vehicle traffic densities.
  • Delay and DPDR: VMaSC-LTE maintains both low delay and high DPDR, whereas flooding and pure cluster-based algorithms exhibit a DPDR–delay trade-off.Among the hybrid architectures, VMaSC-LTE achieves the lowest delay at the cost of infrastructure use.

3) LTE Cost:

LTE usage is controlled through cluster-head and hop-count choices: smaller clusters can raise DPDR toward 100 while increasing the number of clusters and LTE cost.

  • LTE Cost Definition: LTE cost is measured by the number of cluster heads, which depends on maximum hop count and the MEMBERCH constraint.The study varies MEMBERCH from 1 to 10 while treating cluster-head count as infrastructure cost.
  • Cluster Count: Increasing MEMBERCH generally decreases the number of clusters, but cluster count can rise again at high MEMBERCH and hop counts.The increase is linked to higher cluster-head contention, overhead, and disruption when intra-cluster links break.
  • Reliability–Cost Trade-off: DPDR increases up to 100 as MEMBERCH decreases across the evaluated maximum-hop settings.Lower MEMBERCH reduces clustering overhead and IEEE 802.11p contention, improving reliability.
  • Hop Count: Increasing the maximum allowed hops up to 3 increases DPDR at the cost of a slight delay increase.The paper links this improvement to higher clustering stability.
  • Reliability–Cost Trade-off: Higher application reliability requires fewer cluster members, creating more clusters and therefore increasing LTE usage cost.VMaSC-LTE consequently supports adaptive infrastructure use according to reliability requirements.
  • Future Work: Future work will evaluate VMaSC-LTE in urban traffic and extend it with data aggregation and additional vehicle-path information.These extensions are identified as future investigations rather than evaluated results.
Loading 1603.05628v1…