In this post we will try to discuss about a MAC which has more intelligence than what is demanded and possible by the generic implementation of Protocol standards. In Most cases the Protocol standards describes rules and regulation and methodologies to be supported to interoperate and co-exist with different architecture and vendor implementation. It is mostly left open to the level of imagination of the end designer or implementer to use the underlying protocol for purposes not just prescribed by the protocol. Hence exist the design thoughts and principles of Intelligent MAC - the iMAC paradigm.

Before getting in to iMAC principles, it is essential we glance through areas which are building blocks of a Wi-Fi based MAC system where in there is possibility of adding extra intelligence or in some case broad optimization.

1) Channel selection Algorithms

2) Roaming Algorithms

3) Load balancing

4) Rate Control Algorithms

5) Self healing

6) Transmit/Receive engine

Also the latest extension to the 802.11 standards with 802.11r, 802,11k ,802.11v and 802.11z etc has provided a standard, interoperable framework of data communication which can be used for adding value and intelligence to the wireless operation scenarios. Prior adaptations and algorithms were mostly proprietary and had the requirement of infrastructure (Wireless Access point side) and the Mobility ( Wireless Station side) to be from same vendor for custom intelligence and value additions on top of base protocol which is difficult to achieve in most practical scenarios.

Intelligent channel selection

There are several intelligence which can be introduced in the channel selection based on extensive data available over most of the 802.11 extension protocols. This may provide additional value for overall performance of an Access point environment. The typical criteria of number of stations associated to determine the load of an AP is an average criteria of deciding on channel selection.

Ideally a intelligent and adaptive channel selection algorithm can depend on more variables attributed to channel characteristics, collection of data available from protocols and frequent capture of spectral data at pre-configured or triggered periods to run the channel algorithms. Some of the 11k protocol and request report like Beacon, Channel report can be used to create a Dynamic Adaptive and Intelligent Channel selection algorithm.

Roaming algorithms

Roaming is important as most of the Wi-Fi solutions in the end user level typical will necessitate some amount of mobility. May be it is warehouse or hospitals or a godown, roaming always plays an important role. Seamless roaming with minimal data loss can always be a very useful feature in any successful product. The typical optimized roaming algorithm can strengthen its base on following additional intelligence

1) A strong useful effective and non stale data of the infrastructure at any point of time

2) Indication or heuristics of when and how to use the available data.

3) Decision making to use or select the best of the data available for roaming

4 ) Combining the Infrastructure provided information and mapping that with self obtained information at client end to arrive at the right, optimal roaming decision

The 11r and 11k provides many possible intelligence to aid voice quality roaming scenarios. Some of the reports like neighbour and link at the infrastructure side and channel and beacon from the client can be effectively used for highly optimized voice quality support roaming algorithms.

Load balancing

Load balancing is a important component needed to optimize any infrastructure to effectively provide service to it's clients. Load balancing may involve identifying internal and external factors to decide on best possible mechanism to service the associated clients.

Internal factors include

1) No of the clients

2) Type of QOS the application at device may need to support

3) Characteristics of the client ( support for 2.4 /5.0 etc)

External factors include

1) Non 802.11 interference to be tolerated

2) Infrastructure density to be followed

Rate control algorithms

Rate algorithms play a key role in overall performance and throughput availability in any wireless solution. An over aggressive rate climbing algorithm may be impractical due to possibility of higher error in the RF medium and an less aggressive algorithm will result in overall lower net throughput in the system. A adaptive, learning, and "quick to rise", "quick to fall", "stay flat long" seems most optimal to survive any practical conditions giving an above average overall throughput.

The typical decision making factors for the rate algorithms will be

1) Errors in the medium

2) Retries

3) Overall BSS rate histograms

4) AP statistics

4) Spectral data

#80211 #Roamiing #WiFiratecontrol