Algorithm 1: Proposed average common feature extraction technique for DDoS attack detection.

Input: - Dataset tensor $\mathcal{X}\in {\mathbb{R}}^{N_1\times \cdots \times N_R\times M}$ - Multilinear rank $(d_1,\dots ,d_{R+1})$ Output: - Filtered dataset matrix ${\left[\mathcal{X}\right]}_{(R+1)}^{\left[f\right]}\in {\mathbb{R}}^{M\times N}$ Algorithm Steps: 1 Compute the HOSVD of $\mathcal{X}\in {\mathbb{R}}^{N_1\times \cdots \times N_R\times M}$, with multilinear rank $(d_1,\dots ,d_{R+1})$, as in (3) 2 Compute the common feature tensor $\tilde{\mathcal{X}}\in {\mathbb{R}}^{N_1\times \cdots \times N_R\times d_{R+1}}$ as in (4) 3 Compute the average common feature tensor $\overline{\mathcal{X}}\in {\mathbb{R}}^{N_1\times \cdots \times N_R}$ as in (5) 4 Convert $\mathcal{X}$ into the $(R+1)$-th mode unfolding matrix ${\left[\mathbf{X}\right]}_{(R+1)}\in {\mathbb{R}}^{M\times N}$ as in (6) 5 Obtain the weight tensor $\mathcal{C}\in {\mathbb{R}}^{N_1\times \cdots \times N_R}$, whose elements are computed as in (7) to (9) 6 Obtain the concatenated tensors ${\mathcal{C}}^{\mathsf{C}}\in {\mathbb{R}}^{N_1\times \cdots \times N_R\times M}$ and ${\overline{\mathcal{X}}}^{\mathsf{C}}\in {\mathbb{R}}^{N_1\times \cdots \times N_R\times M}$ as in (10) and (11) 7 Compute the Hadamard product between ${\mathcal{C}}^{\mathsf{C}}$ and ${\overline{\mathcal{X}}}^{\mathsf{C}}$ as in (12) 8 Compute the filtered dataset tensor ${\mathcal{X}}^{\left[\mathrm{f}\right]}\in {\mathbb{R}}^{N_1\times \cdots \times N_R\times M}$ as in (13) 9 Convert ${\mathcal{X}}^{\left[\mathrm{f}\right]}$ into the $(R+1)$-th mode unfolding matrix ${\left[\mathcal{X}\right]}_{(R+1)}^{\left[f\right]}\in {\mathbb{R}}^{M\times N}$ as in (14)