 Leaf test codes are secure sine they would not be jammed by jammers.  When few normal users are present, many leaf code tests are wasted since absent users cannot not report jamming.  Optimal height for test codes in the tree is dependent on the number of normal users present, absent users, and jammers.  A subtree behaves according to its leaves: - A subtree is jammed if any of the leaves of the subtree are jammers. - A subtree is absent if all of the leaves of the subtree are absent. - A subtree is normal otherwise.  These designations reflect how the network will react when the root of that subtree is chosen as a test code.  Given there are 2 n total users, of which A are absent, J > 0 are jammers, and N > 0 are normal, The probability of detecting jamming when a test code is chosen at height m is:  The expected number of tests until detection:  The optimal height minimizes the expected number of tests until detection  We performed MATLAB simulations with 20 normal users and 0 to 10 jammers.  Frequency hopping CDMA with 127 channels and 63 hops per bit  Noise across the entire spectrum is 15dB higher than the total signal power.  Power emitted by each jammer equals that emitted by the transmitter.  We simulated 10 tests where the base station transmitted 10,000 6-bit messages  Three scenarios simulated: - Ignore jamming: transmitter uses only one spread spectrum code - Our scheme: network follows our protocol - Best possible: jammers only emit noise  Our scheme delivers almost 100% of packets when there are 5 jammers or fewer and delivers more than 90% of packets between 6 to 10 jammers.  When the jammers have no knowledge of the spread spectrum code used by the system, the network can delivery almost 100% of the packets, but spread spectrum does not mitigate jamming in wireless broadcast networks if the network ignores jammers who have obtained the code. Cross-Layer Jamming Mitigation in Wireless Broadcast Networks Jerry T. Chiang, Yih-Chun Hu Illinois Center for Wireless Systems  Jamming is hard to efficiently prevent in broadcast wireless networks.  Transmitting to all receivers using only one spread spectrum code allows any legitimate receiver to jam.  Transmitting to all different receivers using individual codes is not power efficient. Problem Statement Proposed Solution – Tree Coding Optimization – Test Height Research Results Background  Mobile communication networks are susceptible jamming attacks.  Jamming introduces noise, lowering the signal-to-noise ratio and reducing the probability of successful packet reception.  Jamming can be mitigated using spread spectrum where the transmitter redundantly encodes information using a code allowing the receiver to reject signals that do not come from the transmitter.  Total transmit power is divided between codes in use.  Power efficiency is directly related to number of codes in use. Broader Impact  Our work allows all existing broadcast channels used in spread spectrum wireless access technologies to defend against jamming attacks.  Our work has broad applicability since there already exist a wide variety of spread spectrum wireless access technologies such as IEEE , Bluetooth, IS-95, and CDMA2000. Example: User N 2 holds codes C 2, C 23, C 03, C 07 An example run of our protocol that isolates the effect of jammer, N 5 Disjoint covers shown in blue, detectable codes shown in cyan, and test codes shown in red We propose a power efficient scheme to mitigate jamming in wireless broadcast networks using balanced binary tree:  Each element of the tree corresponds to a spread spectrum code.  Each user is assigned the codes corresponding to a leaf and all its ancestors.  A disjoint cover is a set of spread spectrum codes such that any user can decode using exactly one code in the set.  Each packet sent simultaneously on a disjoint cover and a set of test codes.  The ancestor of a test code in the disjoint cover is called a detectable code.  If any receiver receives packets on a test code but not the corresponding detectable code, the receiver reports J AMMING D ETECTED to the transmitter using his leaf code.  The transmitter stops using the detectable code on which jamming is detected and instead uses its two children codes.  Thus the transmitter always sends packets on the minimal disjoint cover on which no jamming had been previously detected.  Our protocol mitigates jamming by isolating jammers on individual codes while conserving power efficiency by keeping normal users in groups thereby transmitting on fewer spread spectrum codes.