1 Let us first the basic recalls of the~\cite{HLG09} article.
4 The precise the context of video sensor network as represented for instance
5 in figure~\ref{fig:sn}.
9 \includegraphics[scale=0.5]{reseau.png}
10 \caption{SN with 10 sensor}\label{fig:sn}.
15 Let us give a formalisation of such a video network sensor.
16 We start with the flow formalising:
18 The video sensor network is represented as a strongly
19 connected oriented labelled graph.
21 the nodes, in a set $N$ are sensors, links, or the sink.
22 Furthermore, there is an edge from $i$ to $j$ if $i$ can
23 send a message to $j$. The set of all edges is further denoted as
25 This boolean information is stored as a
26 matrix $A=(a_{il})_{i \in N, l \in L}$,
31 1 & \textrm{if $l$ starts with $i$ } \\
32 -1 & \textrm{si $l$ ends width $i$ } \\
33 0 & \textrm{otherwise}
38 Let $V \subset N $ be the set of the video sensors of $N$.
39 Let thus $R_h$, $R_h \geq 0$ be the encoding rate of video sensor $h$, $h \in V$.
40 Let $\eta_{hi}$ be the production rate of the $i$ node, for the $h$ session. More precisely, we have
45 R_h & \textrm{if $i$ is $h$} \\
46 -R_h & \textrm{if $i$ is the sink} \\
47 0 & \textrm{otherwise}
51 We are then left to focus on the flows in this network.
52 Let $x_{hl}$, $x_{hl}\geq 0$, be the flow inside the edge $l$ that
53 issued from the $h$ session and
54 let $y_l = \sum_{h \in V}x_{hl} $ the sum of all the flows inside $l$.
55 Thus, what is produced inside the $i^{th}$ sensor for session $h$
56 is $ \eta_{hi} = \sum_{l \in L }a_{il}x_{hl} $.
59 The encoding power of the $i$ node is $P_{si}$, $P_{si} > 0$.
61 The distortion is bounded $\sigma^2 e^{-\gamma . R_h.P_{sh}^{}2/3} \leq D_h$.
63 The initial energy of the $i$ node is $B_i$.
65 The overall consumed power of the $i$ node is
66 $P_{si}+ P_{ti} + P_{ri}=
67 P_{si}+ \sum_{l \in L}a_{il}^{+}.c^s_l.y_l +
68 \sum_{l \in L} a_{il}^{-}.c^r.y_l \leq q.B_i.
71 The objective is thus to find $R$, $x$, $P_s$ which minimize
72 $q$ under the following set of constraints
74 \item $\sum_{l \in L }a_{il}x_{hl} = \eta_{hi},\forall h \in V, \forall i \in N $
75 \item $ \sum_{h \in V}x_{hl} = y_l,\forall l \in L$
76 \item $\dfrac{\ln(\sigma^2/D_h)}{\gamma.P_{sh}^{2/3}} \leq R_h \forall h \in V$
77 \item \label{itm:q} $P_{si}+ \sum_{l \in L}a_{il}^{+}.c^s_l.y_l +
78 \sum_{l \in L} a_{il}^{-}.c^r.y_l \leq q.B_i, \forall i \in N$
79 \item $x_{hl}\geq0, \forall h \in V, \forall l \in L$
80 \item $R_h \geq 0, \forall h \in V$
81 \item $P_{sh} > 0,\forall h \in V$
85 To achieve a local optimisation, the problem is translated into an
86 equivalent one: find $R$, $x$, $P_s$ which minimize
87 $\sum_{i \in N }q_i^2$ with the same set of constraints, but
88 item \ref{itm:q}, which is replaced by:
90 $$P_{si}+ \sum_{l \in L}a_{il}^{+}.c^s_l.\left( \sum_{h \in V}x_{hl} \right) +
91 \sum_{l \in L} a_{il}^{-}.c^r.\left( \sum_{h \in V}x_{hl} \right) \leq q.B_i, \forall i \in N$$
94 The authors then apply a dual based approach with Lagrange multiplier
95 to solve such a problem.
96 They first introduce dual variables
97 $u_{hi}$, $v_{h}$, $\lambda_{i}$, and $w_l$ for any
98 $h \in V$,$ i \in N$, and $l \in L$.
99 They next replace the objective of reducing $\sum_{i \in N }q_i^2$
100 by the objective of reducing
102 \sum_{i \in N }q_i^2 +
103 \sum_{h \in V, l \in L } \delta.x_{hl}^2
104 + \sum_{h \in V }\delta.R_{h}^2
106 where $\delta$ is a \JFC{ formalisation} factor.
107 This allows indeed to get convex functions whose minimum value is unique.
109 The proposed algorithm iteratively computes the following variables
112 \item $ u_{hi}^{(k+1)} = u_{hi}^{(k)} - \theta^{(k)}. \left(
113 \eta_{hi}^{(k)} - \sum_{l \in L }a_{il}x_{hl}^{(k)} \right) $
115 $v_{h}^{(k+1)}= \max\left\{0,v_{h}^{(k)} - \theta^{(k)}.\left( R_h^{(k)} - \dfrac{\ln(\sigma^2/D_h)}{\gamma.(P_{sh}^{(k)})^{2/3}} \right)\right\}$
118 \lambda_{i}^{(k+1)} = \lambda_{i}^{(k)} - \theta^{(k)}&.&\left(
120 \sum_{l \in L}a_{il}^{+}.c^s_l.\left( \sum_{h \in V}x_{hl}^{(k)} \right) \right. \\
121 && - \left. \sum_{l \in L} a_{il}^{-}.c^r.\left( \sum_{h \in V}x_{hl}^{(k)} \right) - P_{si}^{(k)} \right)
126 $w_l^{(k+1)} = w_l^{(k+1)} + \theta^{(k)}. \left( \sum_{i \in N} a_{il}.q_i^{(k)} \right)$
130 $\theta^{(k)} = \omega / t^{1/2}$
133 $q_i^{(k)} = \arg\min_{q_i>0}
137 \sum_{l \in L } a_{il}w_l^{(k)}-
148 v_h^{(k)}.\dfrac{\ln(\sigma^2/D_h)}{\gamma p ^{2/3}} + \lambda_h^{(k)}p
156 \arg \min_{r \geq 0 }
159 -v_h^{(k)}.r - \sum_{i \in N} u_{hi}^{(k)} \eta_{hi}
168 \sum_{i \in N} \left(
169 \lambda_{i}^{(k)}.(c^s_l.a_{il}^{+} +
176 where the first four elements are dual variable and the last four ones are