1. Managing the performance of multiple radio Multihop ESS Mesh Networks.
March
Managing the performance of multiple radio Multihop ESS Mesh Networks.
Francis daCosta
Meshdynamics
(408)

2. 1 radio ad hoc Mesh Networks
March
1 radio ad hoc Mesh Networks
Severe bandwidth constraints with each hop
Vulnerable to both inter and intra channel interference
Clients mobility adds to the complexity of the problem
Not an enterprise class solution

3. 1 radio ad hoc vs. 2 radio Infrastructure
March
1 radio ad hoc vs radio Infrastructure
Ethernet Link
Ethernet Link AP
0,1 AP
0,1
Bandwidth loss each
hops in network
Bandwidth preserved
at all hops in network RL
1,1 AP
1,1
STA1
STA1
1/2 AP
2,1
STA2 RL
2,1
STA2
1/4
ONE RADIO SYSTEM
Bandwidth halved each level
Routing paths are fixed
Not redundant or re-configurable
Not scaleable or self managing
TWO RADIO SYSTEM
Bandwidth conserved at each level
Flexible routing paths
Redundant and re-configurable
Scaleable and self managing
Routing Path
Alternate Routing Path

4. A two radio Infrastructure Mesh
March
A two radio Infrastructure Mesh
ROOT
ROOT
RELAY
RELAY
ST9
RELAY
ST2
ST8
ST1
ST3
RELAY
ST4
ST5
ST6
Backhaul Up
Backhaul Dn and Access Point to clients

5. Supports Multiple Routing Paths
March
ROOT
ROOT
RELAY
RELAY
ST9
RELAY
ST2
Alternate Path
ST8
ST1
ST3
RELAY
ST4
ST5
ST6
Supports Multiple Routing Paths
Ensures same available Bandwidth at different levels
Self-managed performance Dynamic Load balancing
Backhaul Up
Backhaul Dn and Access Point to clients

6. Meshed ESS is the Wireless Equivalent of 802.1d Switch Stack
March
Meshed ESS is the Wireless Equivalent of 802.1d Switch Stack

7. Hub vs. Switch Topologies: Switch requires similar radios.
March
Hub vs. Switch Topologies: Switch requires similar radios.

8. March
Infrastructure
Hybrid
Mesh
Topologies
Ad Hoc

9. 60Kb Control Layer Embedded Software Upgrade to devices 1
ACG Product offering
March
Monitor
Network
Manage
Settings
60Kb Control Layer Embedded Software Upgrade to devices 1 2
Communications Interface with NMS

10. March
Monitor
Network

11. March
Manage
Settings

12. Backhaul throughput = 50 Backhaul latency = 1.0
March
Backhaul throughput = 50 Backhaul latency = 1.0
ROOT
Low latency network configuration
Low Latency for all nodes
Poor throughput for distant nodes
Throughput is sacrificed for latency
Signal Strength varies inversely with Distance. Devices connect based on Best Local Signal Strength, not best throughput
In the current implementation of the WLAN network depicted, signal strength affects overall network throughput (= 50.0 )

13. Backhaul throughput = 64 Backhaul latency = 1.6
March
Backhaul throughput = 64 Backhaul latency = 1.6
High throughput network configuration
Distant nodes connect through nearer nodes
More hops required for distant nodes
Now latency is sacrificed for throughput
ACG Software control layer enables each AP to make routing decisions that increase the overall throughput. (20% increase to 64)
Higher throughput is achieved at the cost of higher latency. The Average number of hops increases to 1.6.

14. Backhaul throughput = 59 Backhaul latency = 1.2
March
Backhaul throughput = 59 Backhaul latency = 1.2
NMS can “tune” network between extremes
Directive sent to control layer in each device
Devices change associations per config setting.
ACG software layer in AP dynamically reconfigures network to satisfy latency objectives while minimizing throughput degradation

15. Backhaul throughput = 55 Backhaul latency = 1.1
March
Backhaul throughput = 55 Backhaul latency = 1.1
At 37% tradeoff
Backhaul more latency centric
NMS can “tune” network between extremes
Directive sent to control layer in each device
Devices change associations per config setting.
Further incentive (37) reduces the latency to 1.1

16. Backhaul throughput = 50 Backhaul latency = 1.0
March
Backhaul throughput = 50 Backhaul latency = 1.0
At 49% tradeoff
Backhaul at low
latency setting
At 49, the cost of connecting to a parent further removed from the root (in terms of number of hops) is too high.
Thus, the system can be tuned to anything between low latency and high throughput

17. Small footprint control layer also provides: Dynamic Load balancing
March
Congested Backhaul:
Congested Node
Small footprint control layer also provides:
Dynamic Load balancing
Automatic discovery, self healing
Switching, automatic channel allocation
As the load on a node increases, children seek alternate routes based on an increased cost-to-connect value..

18. Small footprint control layer also provides: Dynamic Load balancing
March
Connectivity
Cost = 7
Small footprint control layer also provides:
Dynamic Load balancing
Automatic discovery, self healing
Switching, automatic channel allocation
As the load on a node increases, children seek alternate routes based on an increased cost-to-connect value..

19. Small footprint control layer also provides: Dynamic Load balancing
March 8
Connectivity
Cost = 8
Small footprint control layer also provides:
Dynamic Load balancing
Automatic discovery, self healing
Switching, automatic channel allocation
As the load on a node increases, it encourages its children to find alternate routes.. By progressively increasing the connect cost

20. Multiple Roots for more redundancy, throughput
March
Multiple Roots for more redundancy, throughput
Small footprint control layer also provides:
Dynamic Load balancing
Automatic discovery, self healing
Switching, automatic channel allocation
System supports multiple roots for redundancy and increased bandwidth

21. Small footprint control layer also provides: Dynamic Load balancing
March
Backhaul is
Self healing
Turned Off
Turned Off
Small footprint control layer also provides:
Dynamic Load balancing
Automatic discovery, self healing
Switching, automatic channel allocation
Nodes self configure in case of node failure. System inherently redundant and fail safe.

22. March 13 2004 CoS based Data Flow
Buffer Depletion with Weighted Fair Queuing – QoS ON

23. March

24. March

25. March

26. Implementation of algorithms on Hardware
March
Implementation of algorithms on Hardware

27. March
Network Monitor

28. March
Software
Integration

29. Demonstration on hardware available on request.
March
Demonstration on hardware available on request.
Meshdynamics
1299 Parkmoor Ave, San Jose, CA 95126
Phone: (408)