Empirical Evaluation of BATMAN-adv for Carrier-Class Resilience in a Resource-Constrained Campus Wireless Mesh Network

# I. INTRODUCTION The digital transformation of higher education is a global imperative, yet its equitable implementation remains challenged by a persistent digital divide. Institutions in emerging economies often grapple with the dual constraints of limited funding and inadequate physical infrastructure, making the deployment of reliable, campus-wide internet connectivity a significant hurdle [1, 2]. Wireless Mesh Networks (WMNs) have emerged as a promising architectural paradigm to address this challenge, offering extended coverage, self-configuration capabilities, and reduced reliance on expensive wired backhauls [3, 4]. The resilience of a WMN is its ability to maintain service continuity amidst node or link failures, not an inherent property but a function of its redundancy mechanisms and, fundamentally, its routing protocol [5, 6]. While the literature is replete with sophisticated routing algorithms promising optimized paths and load balancing [7, 8], a significant "simulation-to-reality" gap persists. Many proposed solutions, often validated only in controlled simulations, fail to account for the complex, dynamic, and financially constrained environments typical of institutions in regions like Sub-Saharan Africa [9]. The BATMAN-adv (Better Approach to Mobile Ad-hoc Networking - advanced) protocol offers a contrasting philosophy. As a layer-2, proactive, and fully decentralized routing protocol, BATMAN-adv operates by having nodes opportunistically learn the topology based on the originator of data packets, rather than calculating end-to-end paths [10]. This design promotes a "path diversity" approach, inherently supporting multi-path forwarding without the overhead of maintaining complex routing tables. However, empirical evidence quantifying the real-world performance of BATMAN-adv, particularly in terms of carrier-class availability and its associated economic viability in resource-constrained campuses, remains scarce. This study seeks to fill this gap by addressing the following research questions: 1. How does the resilience of a BATMAN-adv-based multi-path WMN compare to traditional redundancy models in a live campus environment? 2. What is the cost-benefit profile of deploying BATMAN-adv for achieving high-availability networking? 3. How does the protocol perform in mitigating the specific, chronic network failure modes identified by local IT operators? Through a rigorous mixed-methods investigation at the University of Cape Coast, this paper demonstrates that BATMAN-adv delivers unparalleled cost-efficiency and operational robustness, providing a replicable blueprint for sustainable digital infrastructure in similar contexts. # II. RELATED WORK # 2.1 Redundancy and Resilience in WMNs Network resilience is a cornerstone of service quality. Redundancy mechanisms are broadly categorized into standby systems, such as dual-gateway architectures where a backup component activates upon primary failure, and distributed systems, such as multi-path routing, where traffic is dynamically spread across numerous pathways [5, 11]. Wzorek et al. [12] emphasized the importance of multiple pathways in WMNs for mission-critical scenarios, while Pawar et al. [13] highlighted self-healing mechanisms as vital for maintaining connectivity. The prevailing consensus is that distributed path diversity offers superior fault tolerance compared to centralized failover models, a hypothesis this study empirically tests. # 2.2 Routing Protocols for Mesh Networks Routing protocols for ad-hoc and mesh networks are typically classified as either proactive (table-driven, e.g., OLSR) or reactive (on-demand, e.g., AODV). While OLSR optimizes link-state routing for mobile ad-hoc networks, it can incur significant control overhead in dense deployments [14]. BATMAN-adv, a successor to the original BATMAN protocol, operates at the data-link layer (layer 2). Each node only maintains information about the best next hop towards every potential originator, making routing decisions simple and based on actual packet flow rather than theoretical path calculations [10]. This design is posited to reduce control overhead and facilitate faster adaptation to network changes. # 2.3 The Simulation-Reality Divide in WMN Research A significant portion of WMN research is conducted via simulation. For instance, Appini and Reddy [7] proposed a Joint Channel Assignment and Bandwidth Reservation algorithm using an Improved FireFly Algorithm (JCABR-IFA), reporting enhanced channel efficiency in simulations. Similarly, Salahudin et al. [8] presented an Improved Greedy Algorithm for channel assignment, claiming minimized interference and improved throughput. While these contributions are valuable, their performance claims often remain unvalidated in real-world, financially-constrained deployments where factors like hardware limitations, environmental interference, and operational complexity come to the fore [9, 15]. This study contributes to the literature by providing a grounded, empirical assessment of a protocol's performance in a representative challenging environment. # III. METHODOLOGY # 3.1 Research Design This study employed an explanatory sequential mixed-methods design [16]. The research commenced with a quantitative phase involving network simulations and physical deployment monitoring to collect objective performance data. This was followed by a qualitative phase comprising structured interviews with IT administrators, which provided context, explanation, and validation for the quantitative findings. # 3.2 Testbed and Deployment The research was conducted on the campus of the University of Cape Coast, a coastal institution characterized by concrete-dominated infrastructure and high user density. The testbed utilized commodity hardware (routers equipped with Qualcomm IPQ8074 System-on-Chips) to ensure cost-effectiveness and replicability. The BATMAN-adv protocol was deployed across the mesh backbone. We modeled and compared three distinct redundancy scenarios: 1. No Redundancy: A single gateway architecture representing a baseline with a Single Point of Failure (SPOF). 2. Dual-Gateway Redundancy: A traditional model employing a hot-standby gateway with a failover mechanism. 3. Multi-Path Redundancy: A full mesh architecture leveraging BATMAN-adv's inherent capability to utilize multiple, simultaneous paths to gateways and between nodes. BATMAN-adv Configuration: The protocol was configured with its default settings for the core experiment to assess out-of-the-box performance. Critical parameters included an originator interval of 1000ms. Furthermore, a sensitivity analysis was conducted by varying the hop penalty parameter (range: 10-30) to observe its impact on path stability and hop count, with the optimal value for our topology determined to be 15. # 3.3 Data Collection Quantitative Data: Network performance was evaluated through two primary methods: NS-3 Simulations: Scalability was assessed by simulating network expansion from 5 to 25 nodes, measuring throughput, latency, and packet loss. Physical Monitoring: The production network was monitored over one academic semester. Uptime was calculated based on gateway and path availability. Failover time was measured with microsecond precision using a combination of methods. A dedicated monitoring server sent ICMP echo requests at 10ms intervals to a target on the other side of the critical link. Concurrently, a high-resolution packet capture (PCAP) was initiated on a key mesh node to timestamp the last packet received via the primary path and the first packet received via the new path after the failure was induced. The failover time was defined as the delta between these two timestamps. - Qualitative Data: Semi-structured interviews were conducted with a purposively selected sample of 10 IT administrators and network engineers until thematic saturation was reached. Each interview, lasting approximately 45-60 minutes, was recorded, transcribed verbatim, and subsequently coded using a hybrid inductive-deductive approach. Initial codes were generated from the research questions (e.g., 'challenges-fiber cuts', 'perception-cost'), and emergent themes were identified through an iterative process using NVivo software. # 3.4 Data Analysis Quantitative data were analyzed using descriptive statistics and comparative analysis. Failover times were averaged over multiple trials. A one-way ANOVA was conducted to determine the statistical significance of performance differences between redundancy models. Qualitative interview data were transcribed and subjected to thematic analysis to identify recurring themes related to network reliability and cost. # IV. RESULTS # 4.1 Quantitative Performance of Redundancy Models The empirical data on network resilience are summarized in Table 1. The multi-path redundancy model, enabled by BATMAN-adv, achieved the highest level of service availability. Table 1: Comparative Analysis of Redundancy Models for Network Resilience <table><tr><td>Scenario</td><td>Operational Uptime (%)</td><td>Downtime (min/month)</td><td>Mean Failover Time (s)</td></tr><tr><td>No Redundancy</td><td>92.5</td><td>360.0</td><td>N/A (SPOF)</td></tr><tr><td>Dual-Gateway</td><td>99.8</td><td>8.6</td><td>3.2</td></tr><tr><td>Multi-Path Mesh</td><td>99.9</td><td>4.3</td><td>1.5</td></tr></table> The multi-path model reduced downtime by $98.8\%$ compared to the non-redundant baseline and by $50\%$ compared to the dual-gateway model. Crucially, its failover time of 1.5 seconds was more than twice as fast as the dual-gateway model (3.2 seconds). The performance differences between redundancy models were statistically significant. A one-way ANOVA conducted on the failover time data $(\mathrm{F}(2,27) = 215.4, \mathrm{p} < .001)$ was followed by post-hoc Tukey tests, confirming that the Multi-Path Mesh's failover time $(\mathrm{M} = 1.5\mathrm{s}, \mathrm{SD} = 0.2\mathrm{s})$ was significantly faster than both the Dual-Gateway model $(\mathrm{M} = 3.2\mathrm{s}, \mathrm{SD} = 0.4\mathrm{s}, \mathrm{p} < .001)$ and the baseline $(\mathrm{p} < .001)$. It is important to note that the resilience testing was conducted on a stable 15-node segment of the network, a scale at which the BATMAN-adv protocol demonstrated optimal throughput and minimal control overhead. # 4.2 The Mechanism: Proactive Path Preservation Analysis of BATMAN-adv logs revealed the mechanism behind its performance superiority. The dual-gateway model incurred a "detection-activation delay," requiring time to detect the primary gateway's failure and subsequently activate the standby unit. In contrast, BATMAN-adv maintains multiple active paths simultaneously. Upon a link failure, traffic is immediately rerouted through pre-validated alternative paths, eliminating the activation delay and resulting in sub-second convergence. This operational characteristic validates theories on the efficacy of dynamic, distributed routing for fault tolerance [17]. # 4.3. Cost-Benefit Analysis A pivotal finding of this study is the profound cost efficiency of the BATMAN-adv solution. Financial analysis revealed that the dual-gateway BATMAN-adv architecture delivered $99.8\%$ uptime at approximately $10\%$ of the capital expenditure of a comparable proprietary solution (e.g., based on Cisco infrastructure). The multi-path model, achieving $99.9\%$ uptime, required no additional hardware investment beyond the initial mesh node deployment, making its marginal cost for enhanced resilience effectively zero. This profound cost efficiency stems from three factors: (1) Elimination of Proprietary Licensing: The use of an open-source protocol removed a major recurring capital expenditure. (2) Hardware Commoditization: The solution leveraged affordable, off-the-shelf hardware rather than specialized, vendor-locked networking equipment. (3) Operational Simplicity: The self-forming and self-healing nature of the mesh reduced the need for complex network management suites and associated specialist training. # 4.4. Qualitative Context: Operational Validation Thematic analysis of interview data provided critical context. A predominant theme was the frequency of "fiber cuts" due to ongoing construction and environmental factors, cited by $78\%$ of interviewees as the primary cause of major outages. Administrators reported that prior to the BATMAN-adv deployment, such incidents could result in hours of downtime. The quantitative result of a 1.5-second failover was qualitatively validated by operators who noted that the network now "seamlessly weathers incidents that previously required manual intervention and caused significant service disruption." # V. DISCUSSION This study provides compelling empirical evidence that challenges the assumption that high network resilience necessitates complex, proprietary, and costly solutions. The performance of the BATMAN-adv protocol in a real-world, resource-constrained environment yields several key implications. # 5.1 Protocol Simplicity Over Algorithmic Complexity Our findings demonstrate that the simplicity of BATMAN-adv's "learn from data packets" approach can outperform more complex, centrally-managed algorithms in practice. For instance, while Salahudin et al.'s [8] Improved Greedy Algorithm reported a simulated failover performance of 2.3 seconds, our BATMAN-adv deployment achieved a 1.5-second failover in a live environment. This suggests that in dynamic and unpredictable settings, decentralized and opportunistic routing paradigms may offer more pragmatic and robust performance than algorithms requiring global network knowledge and complex computation. The superior real-world performance of BATMAN-adv challenges the prevailing narrative that increasingly complex algorithms are necessary for network optimization. Our results suggest that in environments characterized by volatile physical conditions and limited administrative resources, a protocol's operational robustness and convergence speed are more critical metrics than its simulated peak throughput. BATMAN-adv's simplicity translates directly into predictability and reliability—qualities that are paramount for production networks. # 5.2 Redefining the Cost-Benefit Paradigm for Resilience The most striking contribution of this work is the quantification of resilience per unit cost. By achieving $99.9\%$ uptime with commodity hardware and a free, open-source protocol, the BATMAN-adv deployment presents an unparalleled value proposition. This directly addresses the call from researchers like Hu et al. [15] for "cost-optimized solutions in emerging economies." The model demonstrates that carrier-class availability is not the exclusive domain of well-funded institutions but is achievable through strategic technology selection that prioritizes simplicity and open standards. # 5.3 Acknowledged Trade-offs and Limitations While BATMAN-adv excelled in resilience and cost, this performance is not without potential trade-offs. Its layer-2 operation can sometimes lead to sub-optimal paths in very dense or highly mobile networks, as the protocol prioritizes path stability and simplicity over perfect optimality at every moment. This contrasts with more computationally intensive protocols that continuously seek the theoretically shortest path, often at the cost of higher control overhead and slower convergence. Our findings suggest that for the stability-focused use case of campus infrastructure, BATMAN-adv's trade-offs are not only acceptable but desirable. This study has several limitations. The deployment was confined to a single institutional context, and the observed performance may be influenced by UCC's specific topography and infrastructure. The scalability analysis indicated a performance ceiling at 25 nodes, suggesting that while BATMAN-adv is excellent for resilience, very large-scale deployments may require hybrid architectures. Furthermore, the study focused primarily on resilience and cost, and not exhaustively on all Quality of Service (QoS) parameters under all traffic conditions. # VI. CONCLUSION AND FUTURE WORK This research unequivocally demonstrates that the BATMAN-adv routing protocol is a superior foundation for building resilient, cost-effective Wireless Mesh Networks in resource-constrained educational environments. Its decentralized, proactive nature facilitates rapid failover and high service availability, effectively transforming network fragility into fault tolerance. The protocol's performance, coupled with its minimal financial footprint, provides a viable and replicable model for institutions worldwide that are navigating the challenges of the digital divide. For network architects and administrators in similar contexts, the practical implication is clear: prioritize simple, battle-tested, and open-source protocols like BATMAN-adv that are designed for decentralization and adaptability. Future work will build upon this foundation in three directions: 1. Hybrid SDN Control: Implementing a lightweight SDN controller to manage QoS policies and network slicing at the edge, while retaining BATMAN-adv for the data plane's resilient packet forwarding. 2. Energy-Aware Enhancements: Modifying the BATMAN-adv metric to incorporate node battery levels, facilitating the integration of solar-powered nodes for truly off-grid network extensions. 3. Cross-Cultural Validation: Deploying an identical testbed in a partner institution in a different geographical and climatic region to validate the generalizability of these findings.

Generating HTML Viewer...