Facilitating Adaptive Placement of Management and Control Functions in Converged ICT Systems - Advances in Network-Embedded Management and Applications - page 56

deep

INM

Event Correla tion

Ma deira

TMN

ASA

MANNA

OSI

P2P-ba sed

hierarchy

SNMPv1 +
RMON

OpenFlow

CROMA

FOCALE

SNMPv1

BOSS

Mobile
Agents

Femto SON

fla t A-GAP

Hybrid: mobile
a gents + SNMP

ha ndover ma nagement DNA

STP

centra lized distribution fully distributed

Fig. 1. ICT management and control systems on the distribution and hierarchy design
space.

Several distributed management frameworks make use of highly distributed
function placement, for example, In-Network Management (INM) [3], [4], the
Autonomic Service Architecture (ASA) [13], the CROMA architecture [14],
DNA [15], Madeira [16], Focale [17], and the framework proposed in [18].
These frameworks also vary in the depth of applied hierarchies, for example,
INM and CROMA allow for a flexible stacking of hierarchical levels for
objective management and policy management, respectively, while the
framework in [18] is restricted to three layers and Focale's autonomic
management elements suggest cooperation on a relatively flat hierarchy.
These frameworks follow distribution and hierarchy patterns that are in line
with innovative management and control algorithms, such as handover
management [5], [6] and femtocell self-organizing interference management
in [7] in the context of fixed/mobile convergence, flat management and
control protocols such as A-GAP monitoring [19] and STP, and distributed
event correlation frameworks with hierarchical patterns such as [20].
In wireless sensor networks, although highly distributed in nature,
management frameworks capture virtually all coordinates in the design space
[21]. For instance, the centralized system in [22] performs management
operations purely external to the network, while in [23] management tasks are
performed by cooperating network nodes. The motivation for choosing any of
such approaches is to achieve the best compromise between computation and
communication overheads that are dictated by specific management functions.
Hence, WSNs substantiate the claim that coexistence of different architectural
patterns is essential for flexible function placement.