On the issues of selective jamming in IEEE 802.15.4-based wireless body area networks

Abstract

The beacon-enabled mode of IEEE 802.15.4 provides a Time Division Multiple Access (TDMA) method for low power devices by adopting Guaranteed Time Slots (GTS). GTS communication is a potential target for selective jammers where they perform GTS attacks. In GTS attacks, the adversary selectively picks one of the reserved device slots to corrupt its incoming communication. Considering countermeasures, most existing solutions rely on slot position randomization to distribute the harm of the attack over the other GTS slots. These solutions may be effective in the case of full-slot jamming. However, the introduced GTS attacks consider only TDMA property of GTS communication and ignore other important properties like the protocol behavior and the superframe structure effect on the traffic. Considering these properties while performing GTS attacks exempts the adversary from full-slot jamming and renders the existing solutions with no effect. In this paper, we introduce a new efficient version of GTS attacks that benefits from both the standard behavior and the superframe effect on the periodic traffic to conserve the adversary’s resources for the longest period. Additionally, we provide a solution specially developed to mitigate the harm from this detrimental attack. From extensive simulations conducted in this work, it follows that the attack is economic in terms of jamming duration and jamming packets ratio and the solution is efficient in terms of packet delivery ratio, energy consumption and delay overhead.

Appendix

Considering the two boundaries of an arbitrary turn of the period P, we define three parts constituting this turn;

• (1) The difference between the start of the node slot that follows the first boundary and the boundary itself. We refer to this period as the head.

• (2) The period between the first and last slot start located in the turn. We refer to this period as the body.

• (3) The difference between the second boundary and the last slot start located in the turn. We refer to this period as the tail.

We refer to the period between the first boundary and the start of the node slot that precedes this boundary as the shift.

Figure 11 illustrates these three parts.

Fig. 11

An arbitrary turn of P with the introduced three parts: head, body and tail

Let N be the number of superframes between two successive transmissions affected by the IEEE 802.15.4 MAC layer and let slot and SF be the slot duration and full superframe duration, respectively.

The goal of the Appendix is proving the following:

$$ \begin{array}{@{}rcl@{}} P div_{float} SF \leq N \leq P div_{float} SF + 1 \end{array} $$

Firstly, we have:

$$ N = pre\_delaying + |body| + post\_delaying $$

Where:

• - pre_delaying is a number determining whether there is a delay in the beginning of the turn. This variable takes 0 in the case of delay and 1 otherwise.

• - |body| is the number of superframes in body.

• - post_delaying is the number of deferrals caused by the tail.

• Assuming that shift = 0:

  In this case, it is obvious that head = 0, body = (Pdiv_floatSF) × SF and tail = Pmod_floatSF.

  Thus, pre_delaying = 0 and |body| = Pdiv_floatSF.

  Since Pmod_floatSF < SF, then the second packet arrival can at most be delayed once. Then post_delaying ≤ 1.

  As a result: Pdiv_floatSF ≤ N ≤ Pdiv_floatSF + 1.

• Assuming that shift≠ 0:

  We have

  $$ \begin{array}{@{}rcl@{}} body &=&P - (SF-shift) - (P mod_{float}SF + shift) \\ &=&P-SF+shift- P mod_{float}SF-shift \\ &=&P-SF- P mod_{float}SF \end{array} $$

  Since P = (Pdiv_floatSF) × SF + Pmod_floatSF,then,

  $$ \begin{array}{@{}rcl@{}} body &=& (P div_{float}SF) \times SF+P mod_{float}SF-SF \\ &&\quad -P mod_{float}SF \\ &=&(P div_{float}SF) \times SF-SF \\ body &=& (P div_{float}SF - 1) \times SF \end{array} $$

  Therefore,

  |body| = Pdiv_floatSF − 1

  Since |body| is constant, finding a boundary for N implies finding a boundary for pre_delaying and post_delaying.

  Since shift + head = SF then 0 ≤ shift ≤ SF.

  Consequently, we have two possibilities:

  0 ≤ shift < slot and slot ≤ shift ≤ SF, which implies pre_delaying = 1 and pre_delaying = 0, respectively.

  Thus

  $$ 0 \leq pre\_delaying \leq 1 $$

  On the other hand, we have:

  $$ P mod_{float}SF<SF \ \text{and}\ shift<SF $$

  Then

  $$ P mod_{float}SF+ shift<2 \times SF $$

  Therefore

  $$ tail<2 \times SF $$

  As a consequence, we have three possibilities:

  0 ≤ tail < slot, slot ≤ tail < SF + slot and SF + slot ≤ tail < 2 × SF, which implies post_delaying = 0, post_delaying = 1 and post_delaying = 2.

  Thus

  $$ 0 \leq post\_delaying \leq 2 $$

  Therefore

  $$ \begin{array}{lll} P div_{float}SF - 1 & \leq & pre\_delaying + |body| \\ &&+ post\_delaying \\ & \leq & P div_{float}SF +2 \end{array} $$

  Now, we have to prove that this sum cannot take the inequality boundaries i.e., Pdiv_floatSF − 1 and Pdiv_floatSF + 2.

  Since pre_delaying has 0 and 1 as possible values and post_delaying can take the values 0, 1 and 2, it is sufficient to prove that pre_delaying inevitably takes 0 when post_delaying takes 2 and takes 1 when post_delaying takes 0.

  • Suppose that post_delaying = 2:

    This implies

    $$ tail \geq SF+slot $$

    Which is equivalent to:

    $$ P mod_{float}SF+ shift \geq SF+slot $$

    Thus

    $$ shift \geq SF+slot- P mod_{float}SF $$

    Since

    $$ P mod_{float}SF<SF $$

    Then

    $$ shift>slot $$

    Therefore, the first transmission is undeniably deferred, which means pre_delaying = 0.

  • post_delaying = 0

    This means

    $$ tail<slot $$

    i.e.

    $$ P mod_{float}SF+ shift<slot $$

    Then

    $$ shift<slot $$

    Therefore, the first transmission is not deferred, which means pre_delaying = 1.

    This results on:

    $$ \begin{array}{lll} P div_{float}SF & \leq & pre\_delaying + |body| \\ &&+ post\_delaying \\ & \leq & P div_{float}SF +1 \end{array} $$

    i.e. Pdiv_floatSF ≤ N ≤ Pdiv_floatSF + 1