To ensure the rigor, transparency, and reproducibility of this systematic review, the PRISMA protocol (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) was adopted as the reference framework [31] . This internationally recognized standard structures the different stages of the process, from source identification to final analysis, while ensuring consistency and traceability of methodological decisions. This choice is fully justified in a field as cross-cutting as IoT-enabled smart waste management, which draws on research from engineering sciences, urban planning, and the social sciences. By clarifying inclusion and exclusion criteria, PRISMA helps reduce selection bias and ensures maximum representativeness of relevant studies.

Given the fundamentally interdisciplinary nature of smart waste management, this review relies on four scientific databases selected for the complementarity of their contributions. IEEE Xplore was used for its comprehensive coverage of Internet of Things technologies, sensor architectures, and microcontrollers. ScienceDirect provides essential insights into applied environmental engineering approaches. SpringerLink enriches the corpus with studies combining technology, urban planning, and social sciences, which are particularly relevant for analyzing waste management systems from an integrated urban perspective. Finally, Scopus was chosen for its broad coverage, advanced bibliometric tools, and ability to interconnect publications across diverse disciplines. This combination ensures a balanced coverage of the technical, ecological, and socio-institutional dimensions of the topic, while limiting biases linked to disciplinary silos.

The search strategy was based on a combination of generic and specialized keywords, incorporating synonyms, terminological variants, and expressions commonly used in the field of IoT applied to waste management. The queries were formulated in English, the dominant language of scientific publications on this topic. Boolean operators (AND, OR) were used to cross-reference the main concepts: IoT technologies (“IoT”, “Internet of Things”), waste management (“waste collection”, “garbage monitoring”), logistics optimization (“route optimization”, “smart bins”, “sensor-based garbage”), and social or territorial dynamics (“community engagement”, “developing countries”, “Africa”, “Asia”).

The search period, set between 2016 and 2024, captures recent developments in the field, including the diffusion of low-cost solutions (such as Arduino and ESP32), the emergence of long-range networks (LoRaWAN), and the introduction of decentralized technologies such as blockchain into connected waste management systems.

The search strategy was conducted across four major databases (Scopus, IEEE Xplore, ScienceDirect, and SpringerLink), selected for their broad coverage of peer-reviewed literature in engineering, computer science, and environmental studies. These databases were prioritized to ensure access to the most relevant publications on IoT applications and smart waste management. Although other repositories such as Web of Science or certain regional databases may also contain relevant studies, the chosen sources already provide extensive disciplinary and geographical coverage.

A rigorous set of criteria was defined to ensure scientific relevance, contextual adequacy, and technological consistency with the objectives of this review. Priority was given to empirical research conducted in low-resource countries that effectively implemented IoT devices applied to waste collection. Publications that were purely theoretical, focused on industrialized countries, or addressed related domains such as recycling were automatically excluded. Table 1 summarizes the main inclusion and exclusion criteria used to filter the 329 publications initially identified. These criteria narrowed the final corpus to 36 studies relevant for analysis. This filtering framework ensured analytical coherence while maintaining a diversity of case studies useful for examining the transferability of solutions in low-resource contexts, particularly in sub-Saharan Africa.

The selection process of studies was conducted in accordance with the principles of the PRISMA protocol, ensuring transparency in the steps leading from the initial identification to the constitution of the final corpus. Only publications meeting the defined inclusion criteria (section 2.4) were retained, with a constant emphasis on traceability.

The literature search, carried out between March and May 2025, identified 329 publications from IEEE Xplore, ScienceDirect, SpringerLink, and Scopus. After removing duplicates and performing an initial screening based on titles and abstracts, 137 articles were selected for full-text reading. Of these, 101 were excluded due to lack of contextual relevance, purely theoretical approaches, or no direct connection to IoT-enabled waste collection. The final corpus thus consists of 36 empirical studies. Figure 1 illustrates this process through a simplified PRISMA flow diagram, showing the number of records at each stage and the main reasons for exclusion. This scheme reinforces the methodological robustness of the selection.

The methodological rigor of the 36 included studies was assessed using the Mixed Methods Appraisal Tool [32] across five dimensions: clarity of research questions, adequacy of data collection, appropriateness of study design, risk of bias and validity of measurements, and quality of analysis.

All studies obtained high scores (86% - 97%), indicating a solid and homogeneous evidence base. Even at the lower end of this range, the studies reflected satisfactory design and reporting, excluding any weak evidence. This homogeneity results from the rigorous PRISMA-guided inclusion and exclusion process and reinforces the internal validity of the synthesis, while also providing a benchmark for future comparative reviews on IoT-enabled waste collection.

The analysis of the 36 selected studies was based on an inductive thematic coding approach, allowing the identification of recurring dimensions, differentiating factors, and areas of tension related to the adaptation of IoT technologies in low-resource countries. This method enabled the structuring of extracted data into coherent analytical categories, combining technical, social, economic, and environmental dimensions.

The analysis process was carried out in two phases. First, a coding framework was developed around five main thematic axes, which also served as concrete criteria for assessing transferability: (i) technology, including type of sensors, microcontrollers, and energy compatibility; (ii) infrastructure and connectivity, i.e. network reliability and energy autonomy; (iii) community participation, such as citizen involvement and social acceptability; (iv) operational efficiency, including measurable gains in collection frequency or fuel savings; and (v) cost and scalability, covering deployment and maintenance feasibility. A solution was considered transferable when it achieved satisfactory performance across these dimensions while remaining compatible with the constraints of low-resource African urban settings.

Second, each article was read in full and manually coded using a matrix that cross-referenced these five axes ( Table 2 ). Quantitative data (e.g., frequency of use of a sensor or network protocol) were aggregated to produce statistical visualizations. Qualitative data (narratives, contextual limitations, local successes) were synthesized as comparative observations. This dual reading, both qualitative and interpretative, provided a cross-cutting view of practices in smart waste collection and helped build a typology of approaches adapted to low-resource environments.

Data extraction was carried out using a structured synthesis table specifically designed for this review. This tool centralized key information from each study in a systematic and comparable manner, directly linked to the thematic coding axes. The table was developed in Microsoft Excel, which facilitated sorting, linking variables, and generating statistical graphs. This process proved essential for structuring the comparative analysis and highlighting trends, convergences, and divergences across geographical and technological contexts.

The coding framework was organized along thematic dimensions such as hardware (sensors, microcontrollers), network technologies, user/community participation, governance mechanisms, and contextual implementation factors. While the review included 36 studies, some of them reported multiple occurrences (e.g., different sensors, microcontrollers, or network types), meaning that a single study could contribute to several categories within a dimension. As a result, denominators vary across figures, which prevents totals from exceeding 100% and ensures a transparent representation of the coded distributions.

The annual distribution of publications identified in this review reveals a relatively stable dynamic between 2016 and 2022, with an average ranging from one to four articles per year. A clear inflection point, however, is observable from 2023 onward, marking a turning point in the attention given to the adaptation of IoT technologies to waste management in low-resource countries. This growth is linked to the democratization of low-cost microcontrollers and the wider adoption of low-power wide-area networks (LPWAN), particularly LoRa [8] [20] [48] [50] .

This trend, illustrated in Figure 2 , reflects renewed interest in more accessible technological solutions capable of addressing the structural constraints of precarious urban contexts. The dips observed in 2020 and 2022 may be associated both with the global disruption of research caused by the COVID-19 pandemic and with a stagnation in methodological approaches on this topic, due to the absence of significant renewal in analytical frameworks.

From a geographical perspective, Figure 3 highlights a strong concentration of studies in South Asia, particularly in India, which alone represents a large majority of the corpus, followed by Bangladesh and Pakistan. In Africa, Nigeria emerges as the main contributor, alongside a few West African countries such as Benin and Ghana. By contrast, Francophone Africa, and particularly the Democratic Republic of Congo, is almost absent from the experiments identified. This imbalance in scientific production raises questions about the representativeness of the existing results and limits the scope of the developed models in terms of transferability. It underscores the need to broaden study sites in order to better capture the diversity of contexts and local constraints when reflecting on the implementation of IoT in urban waste management.

The majority of the devices identified rely on a homogeneous architecture centered on ultrasonic sensors, present in nearly 46% of cases ( Figure 4 ). This choice reflects a compromise between cost, energy consumption, and ease of integration, but limits adaptability to complex or highly heterogeneous urban environments. Multi-sensor configurations, still in the minority (28%), introduce gains in precision but remain underexplored, particularly in low-infrastructure settings.

In terms of microcontrollers, Arduino dominates (42%), followed by ESP32 (30%) and Raspberry Pi (17%) ( Figure 5 ). These platforms are favored for their low cost and technical accessibility, but their adoption often remains disconnected from contextual constraints such as energy stability or the need for embedded processing. The use of more advanced microcontrollers, or those specifically designed for constrained environments, remains marginal, highlighting a standardized approach with limited orientation toward frugal engineering or contextual resilience.

Communication protocols play a key role in the viability of these systems. Figure 6 highlights the predominance of Wi-Fi–based systems, used in approximately 27% of the studies analyzed. This frequency of use can be attributed to the ease of integrating Wi-Fi modules into low-cost hardware architectures, particularly via microcontrollers such as Arduino or ESP8266. However, this technical option assumes the existence of a stable network, a condition rarely met in low-resource urban contexts, which raises questions about the actual transferability of these devices.

In parallel, the growing use of LoRa and LoRaWAN networks, present in nearly 20% of cases, reflects an attempt to adapt to connectivity constraints. These long-range, low-power protocols offer significant potential in environments where telecommunication infrastructures are fragmented or absent. Nevertheless, their deployment often remains confined to experimental frameworks, with their long-term viability not systematically demonstrated.

Other communication modes, such as cloud-based solutions, GSM, or RFID, remain marginal, while nearly 30% of the studies do not specify the transmission modalities, thus limiting the critical assessment of technological choices. Taken together, these observations underscore the presence of a technological bias in documented approaches, where local reliability is often relegated to the background in favor of theoretical performance. This tension between technical sophistication and infrastructural realities deserves greater attention in future research.

Figure 7 highlights the limited consideration of end-users in the systems studied. Nearly 60% of the solutions include no form of user interaction, confirming a largely technocentric approach. Interaction through mobile applications, present in 31% of cases, is often limited to notification functions without genuine feedback mechanisms. Only 9% of the studies incorporate direct participation, through co-design tools, acceptability surveys, or gamified mechanisms aimed at strengthening user ownership. Although relevant, these initiatives remain marginal. The widespread absence of community engagement raises questions about the sustainability, maintainability, and local anchoring of the proposed solutions.

The most frequently pursued technological objectives relate to collection monito-ring ( Figure 8 ), improvements in operational efficiency, and route optimization. These priorities reflect a logistics-oriented approach, often at the expense of more systemic issues such as waste sorting or environmental analysis. While pragmatic, this functional concentration tends to reduce the transformative scope of the pro-posed solutions.

At the same time, the identified constraints reveal a persistent tension between technical ambitions and contextual feasibility ( Figure 9 ). Although some soluti-ons claim demonstrated scalability, nearly half of the cases mention economic, technical, or regulatory limitations. Profitability in low-resource settings constitu-tes a recurrent design criterion but is rarely associated with mechanisms for susta-inable support. The integration of renewable energy remains marginal, highligh-ting a gap between sustainability discourse and actual technological choices. These trends point to the need to reorient system design toward architectures that are genuinely adapted to the constraints of low-resource territories.

The cross-analysis of studies reveals a constant finding: IoT solutions are transferable to low-resource countries only if they are deeply redesigned, not merely transposed. The most viable configurations are those that combine frugal engineering with a systemic sobriety logic. This is reflected in the use of low-power open-source electronic components, low-bandwidth long-range protocols such as LoRaWAN, and decentralized architectures that minimize dependence on the cloud [12] [35] [43] [61] . These technical choices, modest in appearance, enable resilience in the face of power outages, weak network coverage, and the absence of structured technical support. They contrast with dominant approaches, often inspired by smart city standards from the Global North, whose sophistication becomes counterproductive in environments where service continuity is a luxury.

Beyond technical considerations, the few projects that incorporated contextual variables into their optimization logics (such as road conditions, topographic accessibility, and overflow frequency) achieved significantly more operational results. In these cases, waste collection was no longer designed solely around distance but prioritized according to the physical and social realities of the field. However, the near-total absence of longitudinal evaluation, social acceptability assessments, or post-deployment monitoring limits the scope of the identified experiments. This gap reflects a recurrent bias in IoT approaches in Africa: innovation too often remains technocentric, ignoring usage dynamics, economic trade-offs, and local capacities for appropriation. Truly transferable projects are those that combine the demand for ingenuity with field pragmatism, envisioning technology as an adjustable lever rather than an end in itself [1] [27] [43] .

The limitations identified in the studies reviewed are not solely the result of inappropriate technological choices but stem from a deeper mismatch between the implicit requirements of the systems designed and the social and material realities of the contexts where they are deployed. Many projects rest on the tacit assumption of a minimum functional infrastructure: stable electricity, continuous network coverage, and trained human resources. Yet, in the peripheral neighborhoods of many African cities, these conditions are rarely met [62] [63] . A single sensor requiring regular recharging or a GSM module dependent on a stable signal can become inoperative within the first weeks. The problem lies not in the complexity of the technology but in its inability to withstand discontinuity, irregularity, and lived realities.

Another striking observation is the institutional isolation in which these projects are embedded. Very few are integrated into municipal or regional waste management policies. Most are the result of ad hoc initiatives led by research consortia or startups, often without lasting connections to local authorities. This lack of anchoring prevents the creation of governance, maintenance, and financing ecosystems capable of sustaining the solution beyond the experimental stage. In the absence of structured dialogue among engineers, policymakers, operators, and citizens, technology becomes an alien body, deployed in a territory that cannot support or repair it. This institutional vacuum is not only a technical obstacle; it also contributes to the invisibilization of communities, who remain excluded from decisions that directly concern them.

In response to the structural limitations identified, some studies reveal forms of innovation that are discreet yet significant. These advances do not lie in the introduction of sophisticated technologies but in the capacity to adapt simple solutions to constrained environments. The recurring combination of minimalist sensors and LoRaWAN networks illustrates this logic. Unlike heavier and unstable protocols in Global South contexts, LoRa provides wide coverage with very low energy consumption, enabling systems that can operate for several months without recharging. The integration of microcontrollers such as ESP32 or Arduino, which are locally available and well documented, reinforces system autonomy while facilitating maintenance by non-specialist technicians. Here, innovation is not about power, but about robustness and continuity.

Other studies have proposed devices that go beyond the technical dimension alone by introducing a form of dialogue between the collection system and inhabitants. This link is established through accessible interfaces, SMS alerts, or community reporting systems. In some cases, sending notifications when bins are full or awarding reward points for punctual reports has helped strengthen user adherence. These approaches suggest that the intelligence of a system lies not only in its algorithms but also in its capacity to recognize citizens as co-actors. Finally, attempts to model waste collection routes contextually, by integrating variables such as road conditions, area usage, or risk levels, demonstrate a fine-grained understanding of the urban fabric. Where standardization fails, these approaches show that it is possible to design systems capable of embracing the shapes, ruptures, and urgencies of a living city.

The lessons drawn from this review suggest that a truly functional IoT system in low-resource contexts cannot be limited to superficial adaptation. It must be redesigned from the ground up, based on both vulnerabilities and often underestimated capacities [64] . The model proposed here does not aim to mimic Western standards reduced to their simplest form. On the contrary, it seeks to compose a sober yet intelligent architecture that combines technical robustness, social accessibility, and sensitivity to local dynamics. As illustrated in Figure 10 , this hybrid architecture integrates both the technological components and the governance mechanisms required to adapt IoT-based waste collection to African urban contexts.

The device is based on a contextual priority index constructed from variables rooted in urban realities: physical accessibility of sites, road quality, frequency of overflows, and vulnerability of serviced areas such as markets, schools, or health centers. This index makes it possible to rank waste collection points according to a relevant order of treatment, not based on abstract metrics but on concrete criteria that directly affect residents’ quality of life. Once this prioritization is established, collection routes are optimized locally within each sector by an algorithm that takes into account terrain discontinuities.

On the hardware side, the system relies on ultra-low-power sensors powered by small solar modules, supported by microcontrollers capable of performing simple local computations. This reduces dependence on constant connectivity. Intelligence is distributed rather than centralized, protecting the system against network outages. On the user side, the interface is designed to work with tools already in common use: basic phones, SMS, and USSD menus. This approach is not about reducing technological ambitions but about cultivating an attentive form of ingenuity that turns constraints into drivers of innovation. In a country such as the DRC, it is less about importing technology than about claiming the right to a better organized city, where digital tools act as mediators rather than barriers.

While this review provides valuable insights into IoT-enabled waste collection in developing contexts, several limitations should be acknowledged. First, the search strategy relied on four major scientific databases, which, although comprehensive, may have excluded relevant studies indexed in other repositories, particularly regional databases. Second, only articles published in English and French were included, introducing a potential language bias and possibly overlooking contributions in other widely spoken languages in Africa and Asia. Third, this review synthesized findings descriptively without conducting a meta-analysis, which limits the ability to quantify effect sizes and statistically compare outcomes across studies.

Despite these limitations, the systematic PRISMA-guided approach, combined with the quality appraisal of included studies, ensures that the evidence base remains methodologically robust and relevant for informing future IoT-enabled waste management initiatives in African cities.

The lessons drawn from this review extend beyond the technological field. They call for a broader reflection on how research is produced and how public policies are formulated in low-resource environments. On the academic side, it has become urgent to move beyond the paradigm of isolated technological demonstrations. Too many initiatives still focus on validating prototypes under ideal conditions, without considering the human, economic, and institutional variables that determine their viability. Research on IoT applied to waste management would gain in relevance if it adopted a transdisciplinary stance, bridging engineering sciences, urban studies, the sociology of practices, and the political economy of infrastructures. Such a shift in approach would also require a rethinking of performance evaluation criteria by integrating indicators of social sustainability, operational resilience, and citizen appropriation.

For public policymakers, this review highlights that useful innovation is not that which comes from outside, but that which is adjusted to the local fabric. An effective policy for modernizing waste management in the African context cannot be reduced to tenders for the import of turnkey technologies. It must begin with the establishment of flexible regulatory frameworks, open to experimentation, and capable of supporting local initiatives. This also involves financing pilot projects rooted in neighborhoods, training technicians capable of maintaining equipment, and fostering stronger articulation between municipalities, private operators, and communities. At the municipal level, it would be relevant to create urban innovation units able to test, adapt, and gradually scale up co-constructed solutions. For behind every uncollected waste item lies less a technical problem than a failure of collective organization. And this is precisely what public policies can address if they build on knowledge drawn from the field.