Mack Daddy LTE MAC Schedulers

The stage is set: operators worldwide have chosen LTE as the overwhelming technology of choice. Indeed, the entire vendor industry is gearing up for LTE roll-out, and with the promise of higher speeds and simpler network architecture, we have a winner here which is enticing to the operator community too.

But as a technology vendor, how do you really win?

Answer: By making adoption of the new technology easier.

Breaking it down into technical terms, the LTE access network is now closer to end users and spectral efficiency and low latency requirements are part of the breakthroughs. But that is not all; the resources on the air are still highly prized possessions, and in the current era of driving down cost-per-bit, managing those over-the-air resources effectively is essential for success.

With all the data packets flowing through it, LTE Medium Access Control (MAC) and the scheduling functions are the central parts of the eNodeB managing the precious resources on the air interface. Schedulers handle the allocation of downlink / uplink (DL/UL) air resources for an optimized system, and the performance of the scheduler has the ability to take system performance to an entirely new (higher) level.

Here is how I see it:

Smart scheduler with an easy interface leads to…

Improved efficiency at managing the radio resources and thus…

Improved user experience!

The scheduler takes care of Quality of Service (QoS) aspects on the access side by employing different algorithms for scheduling the data in UL/DL; an efficient algorithm that enhances the QoS manifolds.

Furthermore, when it comes to the performance of these algorithms, efficiency and ease of integration of new algorithms in to the system are key.

A blanket approach won’t work here: a good algorithm obviously works in high throughput environments and takes care of QoS but it also needs to make sure that all UEs (Ed: user equipment, i.e., end devices) get fair allocation and should be able to tune (adjust) according to real-time traffic need. Better yet, one should have a provision for multiple versions of scheduling algorithms tailor-made for the expected traffic since different traffic classes call for different scheduler algorithms.

A smart version should ensure full-use of capacity, avoiding the “starvation” of low priority users and achieving maximum possible throughput.

Algorithms should take into account the following factors

  1. Fairness in scheduling
  2. Efficient utilization of resources
  3. QoS guarantee (i.e., consideration of channel quality)

For a case in point for #3, consider data in the uplink direction to be a best-effort traffic case; the algorithm needs to look at the sending rate instead of just allocating fixed bandwidth.

It is not enough to simply stop at the basic versions of algorithms – because there is an appetite for more! Amidst the standards-based implementation of all L2 and L3 protocol functions, this is a great area for custom work with huge potential to fine-tune the scheduling process in the UL and DL – thus it is essential to have an open interface to MAC to embed the set of algorithms. The convenience here is to be able to add multiple versions using an open and easy interface toward the MAC and the PHY layers.

Such flexibility enables carriers to understand the nuances of channel conditions and compare the scheduler performance using delays, throughput achieved and packet loss – and work toward the end goal of continually improving user satisfaction (an equation with multiple variables indeed!).

Along with my many colleagues at Continuous Computing, I will be interested in seeing (and helping drive) innovation in this area as commercial LTE deployments become a broader reality. I invite your comments below.

Twitter icon