Best Practices for Engineering and Managing an Althea Network

While there are few hard and fast rules on building a mesh network there are several guidelines to create a network that will maintain the highest level of performance and reliability for your customers.

Networking Best Practices:

Mesh Network Reliability:

• Relays should be connected via point-to-point links only, and never as a client off of a PtmP / sector antenna. PtmP radios have a total “airtime” capacity that it shares among its various clients; connecting a relay as a client on a PtmP radio will create a heavy airtime load on the sector and dramatically reduce the performance of the PtmP radio, degrading service for everyone else on the sector.
• For best reliability relays should ideally be “meshed” together no more than 2 hops apart (3 at max). The more relay nodes you build out without peering them together will create a higher chance for a relay failure to orphan customers should one relay along the chain go down, thus minimizing the benefit of the mesh network.

Multiple Relay Hops and Latency:

In real-world examples, provided a mesh network is using ISP grade equipment (i.e. not constructed with consumer-grade wifi routers) each relay hop typically only adds 1ms of latency, or less. This means that even customers many hops deep shouldn’t end up with noticeably higher latency, and maintain a good quality user experience. For most web activities a few milliseconds of latency won’t impact user experience with the sole exception of gaming. Even still, provided the latency from the gateway to the exit node is within roughly 20ms, customers can be multiple hops deep without having it adversely effect performance even when it comes to latency-sensitive activities like gaming or video call streaming.

Still, it is recommended that networks be engineered so that customers not be much more than 5 hops deep to maintain both ideal latency and price differences between customers at the end of the network and customers close to the gateway.

Cost Management Best Practices:

Relay Depth Considerations on Service Cost:

• In an Althea network it is ideal to not build out customers too many hops deep due to the high cost a customer might pay at the end of many hops. If your relays are getting paid a decent amount (e.g. $0.0.1/gb) if customers end up many hops deep, e.g. 5+ hops, those customers will end up paying a disproportionately higher cost than the rest of the customers on the network. While this might be acceptable if it’s necessary to reach hard-to-reach areas, having a large disparity between the highest and lowest paying customer creates a less-than-equitable system.
  • If it is necessary to build many relay hops to connect hard to reach areas alternative methods can be taken to minimize customers at the end of the line paying disproportionately high costs:
    • Lowering the price-per-gb that relays are paid.
    • Lowering the price per gb at the gateway.

Keeping Customers Monthly Price Stable & Minimizing Service Cost Disparity:

In an Althea network costs are broken down into two categories:

1. Support fee: this is a flat monthly fee that is paid out in equal parts over the course of a 30d period.
2. Price per gb: this is paid out in according to the bandwidth usage of the customer – smaller households or lite internet users end up paying lower fees, while large families and / or heavy bandwidth intensive users end up paying higher fees.

To balance the cost to the customer, minimize the gap between low and high paying customers, and keep customers monthly costs as relatively stable as possible it is advisable to set a higher support fee and lower bandwidth fee. A simple calculator that shows this difference can be found here.

Cost Effectively Connecting Multi-Dwelling Units / Apartment Buildings

For multi-dwelling units and apartment buildings where there are many customers it is ideal not to connect each customer with their own antenna. This would only put a heavy load on a PtmP station / sector antenna, making inefficient use of a sectors full potential performance. Instead it is advisable to connect only one, or ideally two (for redundancy – best case scenario to two separate relays) to a building, and then wire each unit individual via ethernet, or via coaxial MoCA network adapters.