KNOWLEDGE-BASED GROUP POLLING (KB-GP)
Traditionally, probability-based
multiple access networks support QoS by stipulating that the network load be
kept at a very low level. Consequently, network bandwidth is wasted
without any guarantee of a timely response. Attempts at higher-level
protocols to support real time traffic requirements are futile without the
support from the underlying networking protocols. Recognizing this, the
KB-GP method has a medium access control component that provides a simpler,
more effective solution over existing multiple access networks for better QoS
support.
In the KB-GP protocol, each
group of stations in the network is periodically polled for data
transmission. If there is no contention among the stations in the group,
transmission is achieved within an optimum time, in the same order as CSMA with
no collision. If there is contention, a binary resolution scheme is
engaged to identify the contenders for service. The grouping, which can
be adjusted dynamically, is controlled by a high level protocol based on traffic
requirements, classes of services and network conditions. This knowledge
base is utilized to balance the service requirements of each group while
maximizing network utilization, ultimately satisfying system requirements.
At one extreme, each station belongs to its own group,
effectively rendering this configuration the equivalent of traditional
polling. This scenario helps illustrate the robustness of KB-GP to use
the whole bandwidth of the network while maintaining its stability. Furthermore,
the binary resolution scheme reduces the numbers of iteration to identify the
contender to a minimum, thereby guaranteeing a worst-case response time
attending to stations of the group.
In essence, the KB-GP protocol is an
adaptive mechanism in a flexible and robust system architecture combining
system requirements and network conditions to achieve optimum network
performance. At light network load, the KB-GP protocol achieves an
optimal response time in the same order as CSMA. At the extreme case of heavy system
load, the KB-GP method is the same as a simple round-robin polling mechanism
with deterministic and stable outcome while achieving maximum bandwidth
utilization. All this is achieved with a flexible and adaptive framework for
prioritizing network requirements.
C-cation has a two-step strategy
for deploying the KB-GP protocol. First, it will upgrade software of
existing wireless network access components to provide QoS improvements.
Second, it will integrate the KB-GP protocol in new chip design for better
efficiency. C-cation believes that with a large population of deployed
network access products, a software upgrade is a very attractive solution to
support new delay sensitive services and to better utilize bandwidth.
Design and implementation of the knowledge acquired from the traffic on these
networks offers other opportunities and is under study at this time.
At the same time, C-cation has conducted an initial analysis
on the improvements offered by KB-GP over the current the WiMax protocol with
promising results. WiMax, which is being designed to support wireless
metropolitan networks as an alternative broadband access to the telco’s DSL and
the CATV’s cable modem, is considered to be complementary to Wi-Fi for
broadband access. Accordingly, the success of one will help stimulate the
growth of the other. Due to its larger geographical coverage and larger
user terminal population to be supported, WiMax requires a robust QoS mechanism
to be successful. Accordingly, C-cation is exploring venues to apply
KB-GP to WiMax for QoS improvements.