Speaking at a recent Cambridge Wireless Small Cell SIG event, ip.access CTO Nick Johnson outlined some of the significant interoperability challenges with multi-vendor radio networks. He warns of potential roadblocks for LTE and argues that the industry needs to embrace multi-vendor interoperability if it is to achieve optimal technical performance, low cost and short time-to-market. We report his views and ask if enough is being done through plugfests and improved specifications alone.
Multi-vendor mobile networks in the past have been geographically split
It's quite common for mobile network operators to have two or three macrocell suppliers. They allocate geographic regions for each (e.g. North/South or East/West) and sometimes play each other off by moving the boundary up/down between them to reward better performance/lower cost. The only interoperability interface is at the regional boundaries, so there's relatively little need for fully optimised interworking.
Small cells introduce a horizontal split because they sit underneath another vendor's macrocell layer. We already have nationwide multi-vendor radio networks with millions of small cells deployed – it's not a niche.
That's been easier to achieve for 3G, especially when installed indoors for poor coverage. The diminishing participation in 3G Small Cell plugfests suggests that the problems are largely overcome.
The tighter integration of LTE Small Cells
As we roll out 4G small cells (especially outdoors), we'll need to achieve tight co-ordination between the small cell, microcell and macrocell layers – potentially every handover could be between layers. Interoperability is well understood to be critical. This is evident from the growing number of participants and test cases at recent 4G Small Cell plugfests.
Unlike 3G, small cells were designed into LTE from the outset. Full co-ordination between small cell, microcell and macrocell layers is achieved by signalling messages over the X.2 interface. Features such as ABS (Almost Blank Subframes) and eICIC allow the macrocells to avoid or align transmissions to give the small cell layer its share of total radio capacity.
This is perhaps less of an issue for indoor 4G small cells, which don't need such tight co-ordination or integration. Most indoor 4G deployments today only use the S1 interface whereas outdoor 4G small cells are likely to require both S1 and X.2.
Specifications are ambiguous and not fit for purpose
The major issue with current specifications are their ambiguity and extensibility. The X.2 interface allows for many proprietary extensions. There are too many options, allowing developers to assign different units or interpret different meanings for parameters. In Nick's experience, multi-vendor open interfaces work best when kept simple and straightforward.
We can make a comparison of the different sizes of interface specifications versus their success:
- Iub is 1300 pages long versus Iu-h < 100
- Iu-r 1200 pages versus Iu-rh ~30
- X.2 is ~180pages
There are few if any successful examples of multi-vendor Iub or Iu-r, while Iu-h has been deployed with millions of small cells today. Small Cell Forum and NGMN studies have shown that some of the key procedures in the X.2 interface are non-interoperable. MLB (Mobility Load Balancing) is so ambiguous it's not useful or workable. ICIC isn't HetNet interoperable although eICIC is. These specifications are not meeting their goal.
Who's the System Integrator?
A key question is often who is responsible for making the interfaces work in live deployments. The answer is rarely the incumbent vendor and is often expensive for the incomer. This increases the barrier for new entrants and makes it expensive for challengers to gain a foothold in the market. This isn't anything new and has been the case since the earliest days of GSM.
Initiatives such as the Small Cell Forum plugfests are removing the sting and risk for multi-vendor integration. We can expect that SON using standard interfaces should also reduce the burden.
With these in place, the risk of surprises and issues during deployment are much reduced. Some want to go further and have a full certification program while others believe the industry is still at too early a stage for that.
Single Vendor versus Multi-Vendor Radio Networks
Many in the industry, especially smaller operators, may be very happy to go down a single vendor route. It makes it simpler for procurement, operations and training, potentially achieving the best network performance by using proprietary features.
However, those taking a multi-vendor route are able to take advantage of the best-in-industry roadmaps, more competitive pricing and achieve truly optimal performance from best-in-class technology.
The risk is that the industry doesn't aggressively pursue a multi-vendor approach, and by default retains only proprietary schemes. Some may be enticed by Cloud RAN and DAS architectures which use the same vendor specific management, security and signalling interfaces but lock-in entire geographic areas to single vendors.
The Bottom Line
The industry must choose if it truly wants a multi-vendor infrastructure supply chain or not. Adopting Cloud RAN wholesale without truly interoperable interfaces could "turn the clock back" and remove the opportunity to benefit from innovation from small cell vendors.
If so, it must encourage clear and simple interface specifications, support interoperability plugfests and procure equipment based on multi-vendor interworking.
Some delegates felt that a certification scheme (similar to that from the Wi-Fi Alliance) is inevitable and would accelerate success. (Analyst Joe Madden posted on this topic recently), while others believe that the plugfest approach is more appropriate at this stage of industry development.
Based on a talk by Nick Johnson (CTO, ip.access) at Cambridge Wireless Small Cell SIG, London, 1 Oct 2014. Slides available online here
Disclaimer: ip.access is a sponsor of ThinkSmallCell. For more information about ip.access and their products, visit ipaccess.com