LTE (Long-Term Evolution) is a well established and dependable protocol for both fixed and mobile wireless networking. LTE allows for hundreds to thousands of devices to be connected to a single tower, with capacity for high bandwith (up to 100+ Mbps). However, it has a few fundamental flaws that make it challenging for anyone outside of the legacy telecom industry to participate in operating LTE networks, and that vastly decrease the efficiency of the network. Althea’s KeyLTE platform addresses these shortcomings, and helps propell LTE’s usefulness into the future of decentralized connectivity and computing.
To understand the architecture of LTE networks, the following terms are required:
- EPC - Enhanced Packet Core. You can think of this as the ‘central server’ that runs the network. It has a variety of responsibilities, including routing of packets, and enforcing throttling or service disconnection for non-payment.
- ENB - eNodeB, also known as a ‘basestation’. This is the radio that broadcasts a signal that allows devices such as handsets (cell phones) to connect.
- Backhaul - this is the connection between the ENB (which is typically located on a tower or tall building) and the EPC (which is typically very far away, hundreds of miles / kilometers distant). This backhaul connection can be fiber, or a combination of fiber and point-to-point radios.
In a legacy LTE network, the ENBs are distributed around the service areas (on towers and tall buildings), and the EPCs are highly centralized. To illustrate this point, there are currently fewer than 20 EPCs running all of the LTE networks in the United States . That’s extremely centralized considering the number and diversity of subscribers and devices using these networks. In the legacy LTE architecture, all of the traffic from each device has to travel to the EPC.
To illustrate how inefficient this is, let’s say you and a friend are standing next to each other and sharing photos via MMS messages. On a traditional LTE network, each of those photos would have to travel from your phone to the ENB, then from the ENB over the backhaul to the EPC (which is very far away), then the EPC would verify that it is allowed to forward the data (meaning you are an authorized subscriber who has paid their bill), then the EPC would forward the data back along the same backhaul path to the same ENB, and from that ENB to your friend’s phone. So those bits traversed hundreds of miles/kilometers just to ultimately travel a few feet/meters.
Now let’s say in the above scenario that your friend hasn’t paid their phone bill, and they are unauthorized to receive your MMS message. In this case, the first part still applies - your data is sent all the way to the EPC, but then the EPC prevents the data from progressing any further, so your friend doesn’t recieve the photo. However, the backhaul bandwidth was still used up in at least one direction (from ENB to EPC), for no reason. Its like sending a letter across the country only to find out that the recipeient has no mailbox, and the letter shouldn’t have been sent in the first place.
KeyLTE addresses this network inefficiency in a very clever way. The core idea behind KeyLTE is that the centralized EPCs are doing too much. Having all this functionality in centralized server farms is helpful for maintenance, but presents too much of a bottleneck for efficient routing. It also makes scaling up a challenge, as EPC hardware has to be extremely fast and reliable, which means it is very expensive. Decentralized computing is far less expensive, and allows for rapid scaling.
With KeyLTE, the functions of the traditional EPC are divvied up into two groups: things that have to be centralized, and things that can be (or should be!) decentralized. While of course there is more going on behind the scenes, the simplified version of this is that tasks related to billing and user/device authentication do have to be centralized to some degree, and so those tasks will continue to be performed in large datacenters and internet exchanges (IX). Tasks related to packet routing do not need to be centralized, and there are vast efficiency gains by allowing data to travel across decentralized mesh networks in the most efficient path possible (which is determined by path latency - the lowest latency path from device A to device B will get priority, all other factors being equal).
Let’s consider the same photo sharing scenario in a KeyLTE network. You send your friend a photo via MMS message from your phone, which is connected to an ENB. This ENB is connected to a KeyLTE router (which contains some EPC functionality). That KeyLTE router is on-site, just at the base of the tower in a network cabinet, so the latency between the ENB and KeyLTE router is less than 1 ms. The KeyLTE router holds the MMS message in queue, and sends a separate (much smaller) message to the centralized datacenter/IX where the KeyLTE server clusters are running. The centralized servers authenticate the users/devices, and grant approval for the message to be delivered. The KeyLTE router at the base of the tower receiveds this approval, and proceeds to forward the MMS message to your friend’s phone.
So the overall result is the same - the MMS message is successfully sent from your phone to your friend’s phone, and the photo is shared. However, the path by which this transaction has succeeded is far more efficient. Instead of sending the large photo file from phone A, hundreds of miles/kilometers away to an EPC, and then all the way back again to phone B, the bulk of the data just traveled to the tower and back, a tiny fraction of the distance. Also, the long distance backhaul between the tower KeyLTE instance and the IX KeyLTE instance was needed only for a tiny text file, just enough to confirm that the subscribers/devices were authenticated and allowed to transmit data [question for dev: is this auth check sent for every transaction, or is the list of authenticated users cached on the edge core KeyLTE instance?)]. The data itself took up no bandwidth in the backhaul connection.
The above scenario demonstrates the significant efficency gains that the KeyLTE platform introduces in data and packet routing. Couple that with reduced cost for deployments and scaling, decreased deployment lead time, simplified network management, and integrated usage-based billing platform, and the built-in infrastructure for public/private partnerships, and it is clear why KeyLTE is destined to be the future of LTE networks everywhere.