Advances, Systems and Applications
Symbols | Explanation |
---|---|
\(\mathbb {E}()\) | Mean value calculation or function |
\(ta_i\) | task i |
\(Da_i^s\) | Size of data sent by task \(ta_i\) |
\(Da_i^r\) | Size of data received by task \(ta_i\) |
UT, DT | Transmission rate of up-link |
DT | Transmission rate of down-link |
\(Cap_{Lo}\) | Computational capacity of UE |
\(Cap_{l}\) | Computational capacity of VM l |
\(Lat_i^{ul},Lat_i^s, Lat_i^{dl},Lat_i^{UE}\) | Latency of task \(ta_i\) from up-link channel, from MEC host side, from down-link channel, and from UE respectively. |
\(\mathcal {T}_i^{U}, \mathcal {T}_i^s,\mathcal {T}_i^{D},\mathcal {T}_i^{UE}\) | Finishing time of task \(ta_i\) on up-link channel, MEC host, down-link channel, and UE |
\(Av_i^{U},Av_i^s,Av_i^{D},Av_i^{UE}\) | For specific task \(ta_i\), the available time of up-link channel, MEC host, down-link channel, and UE respectively |
\(Pol_{1:n}\) | Offloading policies for task set including \(ta_i...ta_n\) |
\(\mathcal {T}_i,\rho (\mathcal {T})\) | A learning task and distribution of learning tasks |
\(s_i, a_i, r_i\) | the i-th state, i-th action, and i-th reward of an MDP |
\(\pi (a|s;\theta )\) | Offloading policy model |
\(v(s;\theta )\) | Value function |
\(\tau _{\pi }\) | Trajectories sampled via policy model \(\pi\). |
\(\mathcal {F}_{en},\mathcal {F}_{de}\) | Encoder functions and decoder function |
\(e_i,d_i\) | Encoder output and decoder output at time step i |
\(c_i\) | Context vector at decoding step i |
\(\hat{A}_t\) | Advantage function value |
\(Up(\theta ,\mathcal {T}_i)\) | Learning optimizer function (e.g., Adam) |