With all the discussion around AI, this question arises: Why do I need an on-site assessment when the vendor says I can use an automated predictive tool? Is it true that we don’t need an on-site assessment? Is a predictive tool good enough? How accurate is AI? Let’s dig into this topic to help you make the proper decision for your environment.
The Importance of On-Site Assessments
When implementing a new upgrade, some people might consider skipping the site survey and just try to model using a predictive tool. However, on-site assessments play a crucial role in creating accurate and reliable network designs. There are several methods to conduct these assessments.
Passive Site Survey
A passive site survey involves a person coming to the fully furnished site—or as close to it as possible—and walking every area of the building. The other primary requirement is that the APs must be installed in the building. To ensure comprehensive coverage, the person must walk several paths, including perimeters and static passes.
AP-on-a-Stick
Another method is “AP-on-a-Stick.” Each person does this slightly differently. Some start with a passive walkthrough to collect background noise, then bring out the hardware and use temporary mounting to simulate the placement of the APs. Others start by placing the demo AP and measuring based on that.
After the surveyor places the demo AP where you expect it to go, they walk reasonably within the expected coverage area, capturing the edges of the coverage cell. Before placing another AP, check if it covers as expected when you layer all these APs in whatever program you use (preferably Sidos). An experienced engineer will adjust the power and placement to create a good design.
An experienced Wi-Fi engineer using a program will then be able to accurately determine the loss of each wall or furniture. No matter how good a program is, I have yet to find one that accurately models reflections. Some programs attempt to do this but often fall short.
The Value and Limitations of Predictive Surveys
Predictive surveys are valuable tools in designing wireless networks, especially when on-site assessments are not feasible. They allow engineers to create a model of the environment and plan the network accordingly. They also make a great budgetary list of items that will be needed for the new deployment. However, they are only as good as the assumptions made during the modeling process.
The Role of Assumptions
Predictive surveys rely on assumptions about the environment, such as wall materials, thickness, shelving, and furniture placement—factors that can significantly affect signal propagation. If these assumptions are inaccurate, the predictive model may not reflect the real-world performance of the network.
Improving Predictive Surveys with On-Site Insights
To enhance the accuracy of predictive surveys, incorporating on-site insights is essential.
Gathering Detailed Information
By involving someone on-site to provide detailed information about the building—such as wall types (drywall, plaster, brick, concrete) and layouts—you can improve the predictive model. While this approach still involves educated guesses about material attenuation, having on-site data allows for more informed assumptions.
If you guess at material types without ever going on-site, your assumptions become even more uncertain. Incorporating on-site insights reduces guesswork and helps create a predictive model that more closely aligns with reality.
The Hybrid Approach: Enhancing Predictive Models with Surveys
An effective strategy to bridge the gap between predictive surveys and on-site assessments is the hybrid approach. This method combines the strengths of both techniques to create a more accurate and reliable network design.
How It Works
In the hybrid approach, you conduct an on-site survey to collect essential data, which is then used to teach and refine the predictive model. Tools like Sidos AI can help in this process by providing a more accurate predictive model based on the collected data.
- Conduct a Survey: Use Sidos Wave to capture essential data in specific areas.
- Teach the Model: The collected data helps the model understand the specific characteristics of your environment, including wall attenuations and existing AP placements.
- Plug-in Missing Information: Manually input details that the model couldn’t figure out due to insufficient data, such as certain wall types or obstructions.
Conclusion
While predictive surveys and AI tools offer exciting possibilities, they are most effective when combined with real-world data. Predictive models are valuable but are only as good as the assumptions they are based on. By incorporating on-site insights and using tools like Sidos AI to build and refine the model, you can reduce guesswork and create more accurate network designs.
The hybrid approach—blending human expertise with AI innovation—is shaping the future of network planning. By staying informed and leveraging these advancements, you can make the best decisions for your environment, ensuring reliable and efficient network performance.
