Introducing ElasticMIMO

An ElasticMIMO system includes the following:

• A solution that supports two Wi-Fi modems concurrently.
• A solution that may operate concurrently at two different Wi-Fi bands – e.g 2.4GHz and 5GHz.
• A solution that can dynamically allocate, at runtime, its antenna system between the two concurrent modems.
• A decision mechanism optimizing at runtime and profiling the most adequate antenna configuration to optimize overall system performance.

Celeno’s CL8000 family introduces an ElasticMIMO Wi-Fi 6 (802.11ax) system that may operate dynamically across the following configurations (runtime changeable):

ElasticMIMO solution architecture:

Different Wi-Fi Bands with Different Characteristics

Wi-Fi utilizes license exempt bands, globally available in the 2.4GHz, 5GHz and expected soon in the 6GHz bands, with 802.11ax. These bands differ in multiple aspects:

1. Available bandwidth –
   1. 2.4GHz spans across ~80MHz (in some countries regulation varies), resulting with 3 non-overlapping 20MHz channels, or 2 non-overlapping 40MHz channels.
   2. 5GHz spans across ~680MHz (in some countries regulation varies) out of which some parts are restricted for use, and some parts are shared with radar systems (mandating Dynamic Frequency Selection).
   3. 6GHz spans across ~1GHz (in Europe would be less) with different sharing schemes with legacy deployed equipment in this band (e.g. point-to-point links).
2. Legacy systems –
   1. 2.4GHz is extensively used by legacy standards (11b, 11g, 11n) operating at 20 / 40 MHz channels. Wi-Fi 6 (11ax) uses this band as well.
   2. 5GHz is extensively used by 11ac, 11n, and 11a solutions. Operating mostly in 80 or 40MHz channels. Wi-Fi 6 (11ax) is expected to use this band as well.
      
      
3. Physical characteristics –
   
   

   Radio waves attenuate while propagating in space, the attenuation is not only a factor of the distance, but also of frequency:

   

   1. Where d is the distance over which the signal propagates, c is the speed of light and f is the signal frequency. This means that higher frequencies decay faster, and for Wi-Fi, it would mean that under same transmission power and distance, 5GHz would decay twice as much as a 2.4GHz signal would, which means that 2.4GHz coverage is better than that of a 5GHz signal.
   2. At Wi-Fi frequency range, physics dictate that higher frequencies decay faster when penetrating through solid objects/physical obstructions such as walls, metal frames etc. Thus, in some scenarios across multi-rooms and multi-floors, 2.4GHz may be able to establish a link while 5 or 6 GHz would not.

Summarizing the above, it is likely for 5 and 6 GHz bands to be able to drive more capacity, due to the likelihood of operating at wider channel bandwidth. On the other hand, 2.4 GHz propagates better over range and through obstacles than 5 GHz or 6 GHz.

Device Types & Device Proliferation

Today’s networks serve a multitude of device types – from mobile to portable, IoT, video-savvy clients, audio, gaming and more.

Wi-Fi home networks serve on average over 10 devices. This number is expected to more than double in the next 4 years, and possibly sooner. A key driver of this is Wi-Fi 6 (802.11ax). Wi-Fi 6 introduces more techniques for dense environment networks with features that are improving the user experience of bandwidth savvy applications like video and gaming(such applications favoring 5 or 6 GHz for the reasons explained above). Wi-Fi 6 also introduces techniques favoring IoT type of devices with narrow bandwidth per device, extended range, and better battery efficiency (such applications favoring 2.4GHz for better range under given transmit power).

While Wi-Fi 5 (802.11ac) applies to 5GHz only, Wi-Fi 6 addresses 2.4GHz, 5GHz and soon also 6GHz. Determining which band will prove to be more important and worthwhile for the services and connectivity experience in a few years’ time is not trivial. Service Providers and OEMs need to design their hardware today, hoping to meet the requirements of future hardware solutions for service and device proliferation that are almost impossible to predict.

Design Cost

Increasing the MIMO dimensions of each Wi-Fi interface appears to cope with unknown future services or device load in each band. However, increasing the MIMO order of a communication system does not linearly add up and is a decreasing logarithmic function to the link budget contribution. Thus, determining to what extent the system should be expanded is a question of diminishing returns, considering that each MIMO chain not only complicates and makes the silicon more expensive but also requires RF PA, LNA, Switch and antenna thereby increasing system costs, consuming more power and generating more heat that needs to be dissipated.

Design Form Factor

Increasing the MIMO dimensions of each Wi-Fi interface not only increases the unit costs and power consumption/dissipation challenge – it may also increase the minimal form factor if not designed carefully.

The antenna system at each band requires proper separation from each other in order to enable the desired diversity gain; common engineering practice is to design the antenna system with at least half wave length (λ/2) separation, which translates to ~5-6cm in 5GHz and ~12cm in 2.4GHz.

Additionally, if silicon implementation mandates external memories, or multiple silicon solutions (e.g. for base-band and RFIC) and consumes more power, the result can be increased layout and form factor of the Wi-Fi solution that may fail to fit the target enclosure and unit deployment model (e.g. failing to fit a power plug extender form factor).

Multi AP – Design for Geo-Capacity

The increased demand for whole home capacity consumption by multiple concurrent devices that are spread across the range makes solutions aiming for scalable geo-capacity favorable. For Wi-Fi, managing a multi-AP network in the form of extenders, repeaters, mesh-nodes is a growingly adopted topology.

Having multiple APs that are serving the entire covered area, with the possibility of different backhaul needs, and the guarantee of different front-haul needs and services, complicates a design of a single piece of easily installed hardware. This is particularly important in consideration of the end-user location preference, while providing the performance benefits and coverage expected.

Nevertheless, for a multi-AP solution, the costs of each node becomes a critical factor for the viability and it allows engineering of cost and form-factor conscious devices that can meet the business case and deployment scenarios desired.

Network Needs Vary Between Homes and Over Time

In each environment (e.g. in each type of home), the location of the AP installed vs. the actual connectivity hot zones is likely to be different (hot zones are considered areas where clients reside and require connectivity and services).

The following diagram illustrates a home environment where the bottom left room requires more capacity (red color) compared to other zones in the home that requires less capacity (blue and yellow color).

Furthermore, in most homes network needs are likely to vary over time. The following diagram illustrates an example of network hot zones through evening / weekends where most of the home residents are present and active, and a second diagram shows daytime/work hours while the home residents are at work/school and the network aiming to serve roaming clients on the go.

Network hot-zones during evening/weekend

Network hot-zones during daytime/work hours

Use case 1 – Capacity Centric, AP serving 11ax and legacy clients

In this scenario, the network is serving 4 Wi-Fi 6 (11ax) clients, each of these is a 2x2 client (a common configuration for mobile 11ax clients) along with 2 legacy clients (marked in the image above with “L”), the clients are spread across the home and are operating concurrently in downlink (AP transmitting to clients).

Rigid MIMO System: Would aim to utilize both interfaces and load balance to best practice – steering the legacy clients to 2.4GHz (as it is a narrower bandwidth) and servings the four 11ax STA with DL MU-MIMO over 5GHz (80MHz channel).

ElasticMIMO System: The optimization algorithm allocated 6 antennas to the 5GHz and 2 antennas to the 2.4GHz, while serving the legacy clients on the 2.4GHz and the four 11ax clients on 5GHz with DL MU-MIMO (80MHz channel).

Comparing Results:

Compared results – four 11ax clients (2x2) & two legacy clients (downlink)

As can be seen, by allocating dynamically 6 antennas to the 5GHz, the capacity of the interface almost doubles with minimal impact on the 2.4GHz legacy client performance.

Running the simulation for the same clients in the uplink direction yields very similar benefits as illustrated in the following result chart:

Compared results – four 11ax clients (2x2) & two legacy clients (uplink)

Use case 2 – Coverage Centric, AP serving Homespot / Hotspot

In this scenario the network is serving 2 Wi-Fi 6 (11ax) set-top-boxes that are active in the home, each consuming dual UHD streams (~100Mbps each STB), while serving additional clients roaming on the street. Roaming clients are spread over a distance, at a pathloss of up to 110-115dB from the AP.

Rigid MIMO System: Would aim to utilize both interfaces and load balance to best practice – serving the 11ax STB on 5GHz, serving the roaming clients on the street on 2.4GHz (4x4). Some of the clients are failing to establish a link.

ElasticMIMO System: The optimization algorithm allocated 6 antennas to the 2.4GHz and 2 antennas to the 5GHz, having enough capacity to easily drive 100Mbps to each of the STBs, while serving the roaming clients much better on the 2.4GHz (6x6).

The performance to the two 802.11ax video devices that require 8K or dual 4K UHD streams (total of 100Mbps per device) at the 5GHz band is the same for both systems. However, with 6 antennas allocated to the 2.4GHz band, the performance of the devices connecting from a remote location is dramatically improved – no devices are out of service and the total performance of those devices is greatly improved.

The additional link budget of the 6 antenna system would also contribute to a much more resilient link, improving the fading-mitigation of the devices connecting from a far location.

Nonetheless, the simulation does not represent this important benefit as the performance of the 4x4 system would likely be reduced due to the high PER caused by the fading.

Use case 3 – Multi Services – STB & Mix of Mobile Clients

In this scenario, the network is serving a mix of dedicated video clients and a multitude of data clients spread across the home. The network is serving four concurrent Wi-Fi 6 (11ax) set-top-boxes, each of 4x4 configuration, consuming all together ~ 250Mbps of video service, along with 2x2 Wi-Fi 6 clients that are concurrently active.

Video service to Set-Top-Box is of higher priority and should be addressed first before serving the data clients. Furthermore, considering the distribution of the STB across the home and the traffic pattern nature of video, the assumption is that either SU-MIMO or SU-OFDMA, and not MU-MIMO, would be used with the STB.

Rigid MIMO: Serving the STB (SU/OFDMA) + 11ax clients (MU) on 5GHz 4x4, and the legacy clients over 2.4GHz.

ElasticMIMO: Algorithm allocated 6 chains to 5GHz serving the STB (SU / OFDMA) + 11ax clients (MU) and 2 chains to 2.4GHz for Legacy clients.

Comparing Results:
The following chart indicates the amount of airtime % required to serve the four 4x4 set-top-boxes with 250Mbps. The x-axis values represent the set-top-boxes’ average path-loss attenuation (in dB) and the y-axis values represent the required airtime %. Any value higher than 100% indicates that there is not enough airtime to serve the 250Mbps to the 4 STB:

Utilizing the remaining airtime to serve the 11ax data clients yields the following result:

Analyzing achievable performance of both systems yields the following observations:

Serving 4 concurrent UHD along with 1Gbps of wireless service –

• A rigid MIMO system fails to achieve the use-case even at short range (50dB path loss, which is in the same room as the AP)
• An ElasticMIMO system can meet the use-case up to ~78dB path loss (which is similar to the expected attenuation between two adjacent rooms)

Serving 4 concurrent UHD along with 500Mbps of wireless service –

• A rigid MIMO system can meet the use-case up to ~77dB path loss
• An ElasticMIMO system can meet the use-case up to 90dB path loss

Use case 4 – Multi AP, Backhaul Considerations

For multi-AP wireless fabric, we have analyzed wireless link performance, whereas for a rigid MIMO system it is a fixed 4x4 to a 4x4 link. For ElasticMIMO system both the root-AP and the extender have their radio elastic.

Links are assumed possible at either 5GHz or 2.4GHz, while at 5GHz operation is assumed using 80MHz channel bandwidth, while using 40MHz in 2.4GHz.

Rigid MIMO: Backhaul link limited to 4x4-to-4x4 with 4SS at either 5GHz or 2.4GHz

Elastic MIMO: Backhaul link may reach up to 6x6-to-6x6 with 6SS at either 5GHz or 2.4GHz

Compared Results:
On 5GHz with 80MHz Channel BW

On 2.4GHz with 40MHz Channel BW

OBSERVATIONS

For 1Gbps of whole home service:

• RigidMIMO: A 4x4-to-4x4 link can deliver 1Gbps of throughput up to 66dB of path-loss, which is similar to the signal attenuation through a single wall (i.e. the next room). Typically it is not expected that an extender will be placed in such proximity (it will more likely be placed two walls away, or a floor up/down, with path-loss of above 80dB).
• ElasticMIMO: Allows both nodes to use 6x6 antennas in either band and can carry the 1Gbps of service over 80MHz channel to the desired location as it can reach this performance up to 84dB path-loss.

For 500Mbps of whole-home service:

• RigidMIMO: A 4x4-to-4x4 link would require 78% of airtime at 80MHz channel bandwidth to carry 500Mbps to the desired extender location (83dB path loss). Thus, under repeater mode it does not have sufficient airtime remaining to serve extended clients (only 22% of airtime remained for service). At 2.4GHz with 40MHz a 4x4 link can’t carry 500Mbps at 83dB path loss.
  
  Furthermore, if the 4x4-to-4x4 link is dedicated for backhaul (e.g. as in tri-band configuration) there is not enough airtime to serve two extenders concurrently both with 500Mbps.
  
  
• ElasticMIMO: Up to 6x6-to-6x6 for the backhaul link at either band. In 5GHz only 47% of airtime is required to carry the 500Mbps to the target extender location, having enough time to accomplish both extender scenario– having enough airtime to serve also front-haul connected clients with 53% of airtime remaining (this would be the scenario of dual band devices); or be able to backhaul in full 500Mbps to two extenders (e.g. in a tri-band scenario when the link is used only to interconnect between the nodes).
  
  At 2.4GHz ElasticMIMO can carry 500Mbps up to 85dB of path loss, making 2.4GHz a viable backhaul option (not possible for 4x4).